When scripting an object tag for the param tags, IE reports all param tags of all object tags as children of all object tags.

Test page Workaround is not included
Reported by: Bart Melton.

When a file upload field has an incorrectly specified file, browsers respond in different ways.

IE solves the problem by preventing form submission.
This approach is reasonably sensible, however it would be much better if IE popped up a warning dialog saying "File not found".

Firefox takes a totally different approach; it lets the server sort out an appropriate response.
It means I have to code the server for this use case (good practice anyhow) and I can put up my own warning messages.

In summary, the Firefox approach needs a bit more coding but gives me complete control over the page. IE means well, but leaves my users confused and unable to leave the page.

Test page Workaround is not included
Reported by: Duncan Simey.

When border-collapse: collapse; is applied to a table with borders and a hidden cell, all browsers have issues:

Firefox hides border in the table entirely if all the cells in the row are hidden but the row element is visible. This can be resolved by hiding the row element.

Internet Explorer renders the border of the row that has all it's cells hidden. This can be resolved by hiding the row element.

Opera doesn't render all of the top border of a table if any of the top row cells is hidden and there is atleast one visible cell on the row.

Safari and iCab don't render the td border at all if a cell is hidden.

Test page Workaround is included
Reported by: Merri.

When using relatively positioned floating elements inside a container, which has an overflow other than visible, the elements behave as if the container has overflow: visible. This occurs in (at least) Internet explorer 6 and 7.

Test page Workaround is not included
Reported by: Viktor Kojouharov.

Internet Explorer crashes when a DOM tree traversing JS function is run on an element with display: none; position: absolute that has child elements.

The crash occurs when assigning the result of a unsuccessful document.getElementById() call to a property of one of the children in the hidden element.

Test page Workaround is not included
Reported by: Steffen M�ller.

I first noticed the problem when I tried to use a PNG+alpha background in an element that contains an hyperlink (see test page). The hyperlink would not work in certain area, which I later identified as non transparent areas of the background image.

It seems as if the filter "steals" mouse events in non transparent areas, regardless of its z-order (no click, no cursor shape change).

Workaround: add the filter to an element with position: static, and there must be a position: relative child element.

Test page Workaround is not included
Reported by: Herve Masson.

The getAttribute() method will return "null" in Internet Explorer when trying to get the for attribute of a label element.

Solution: use attributes['for'].nodeValue.

Test page Workaround is included
Reported by: Laurent Muchacho.

If you define a function as a member of the window object in IE7, the keyword this obviously refers to the wrong object:

window.myFunction = function() {
   return (this === window);
}

A call to myfunction() in IE7 would incorrecty yield false. This problem also affects exception handling. Predefined event-handlers like onload are not affected.

Test page Workaround is included
Reported by: Arnold Konrad.

IE6 throws errors when an element is cloned from one frame and appended to another. This includes pop-ups.

The test page shows two frames and buttons which trigger various combinations of cloning from (hidden) elements in a frame and appending to another.

In Firefox 1.5.09, all permutations work. In Safari 1.3.2, all permutations work at some times, but silent crashes occur at others

IN IE6win, an unappended clone can be held in a variable in another frame (eg. its parent), but it must be appended to the frame from whence it came.

I think this means that DOMloading in pop-ups must be created afresh OR cloned from elements in the pop-up itself.

Test page Workaround is not included
Reported by: Andy Taylor.

Attempting to set a generated checkbox element to a checked status fails in IEwin6:it doesn't matter whether setAttribute is called before or after appending.

Don't have a workaround besides using el.checked=true after appending (which is precisely want I wanted to avoid!)

Same problem doesn't affect disabled attribute. PS. see also testpage elements for select.selected and select.size

Test page Workaround is not included
Reported by: Andy Taylor.

If you clone a table row element (tr) by means of cloneNode()in IE it loses its table row "identity". This means that with the cloned row cells[], for instance, is no longer available.

Test page Workaround is not included
Reported by: Rein Groot.

Repositioned content via JavaScript coupled with floats is triggering a variety of bugs in various browsers.

Test page Workaround is not included
Reported by: Chris Hester.

Changing the status of a checkbox less then a second after it's been changed doesn't change the status again. Try clicking the checkbox twice in rapid succession.

Test page Workaround is not included
Reported by: Velko Georgiev.

Internet Explorer 5, 6 & 7 fail to always render the background colours of repeated divs when you scroll down.

Test page Workaround is included
Reported by: Chris Hester.

Placing a right-floated element in a left-positioned absolutely-positioned element causes the following behavior:

• Opera and iCab display correctly (I think); the absolutely-positioned element is sized to fit the float or other nodes of the element, whichever is wider
• Gecko stretches the absolutely-positioned element to the maximum width allowed by the containing block
• KHTML stretches the absolutely-positioned element to the sum of the width of the float and the other child nodes
• (ppknote: IE behaves as Firefox)

My understanding of the box model spec is that Opera and iCab's behavior is correct; if anyone can corroborate this it would be greatly appreciated.

Test page Workaround is not included
Reported by: Josh Fremer.

If your document's body uses a percentage value as a left or right margin, then position:relative (even without left/top/bottom/right declared), floated elements inside a containing block with a defined width, will not reposition accurately (or at all) as the page is resized. Reloading the page after resizing fixes the problem - just like a Netscape 4 bug.

Test page Workaround is not included
Reported by: David Bloom.

Setting border: 2px inset on a text input element (INPUT.text, INPUT.password, TEXTAREA) to return to its default border after having changed its border style and color leaves the left border's previous color intact.

Test page Workaround is included
Reported by: Jason Brunette.

IE fails to fully render the background and borders of a div containing floats when the browser window is narrowed.

Test page Workaround is included
Reported by: Chris Hester.

In IE6win, an empty div with a css height less than the prevailing font-size (1em) will display with a height of 1em.

The test page provides 2 workarounds.

Test page Workaround is included
Reported by: Henry Francis.

IE can drop a floated column with a percentual width when the browser window is narrowed.

ppknote: Explorer Mac crashes when you resize the test page.

Test page Workaround is included
Reported by: Chris Hester.

Negatively margined (or indented) elements get cropped if their parent elements have layout (hasLayout == true).

(ppknote: IE 7 crops the right margin of the first example, but not the left. IE Mac ignores the margin-right: -10px in the first example.)

Test page Workaround is not included
Reported by: Margaux.

IE 6 returns "unknown" as the type for the getAttributeNode function, instead of "undefined", making object detection more difficult. This happens for nodes in an Msxml2.XMLHTTP object, but not for normal HTML.

Test page Workaround is not included
Reported by: Andy Harrison.

If a parent element is overflow:hidden, any of its children or grandchildren should be invisible if they fall in the overflow area.

In IE 6 and 7 beta 2, if a child has position:relative, its contents are totally visible although the parent is overflow:hidden

Test page Workaround is not included
Reported by: thinsoldier.

The use of italics on long text blocks makes the containing DIV wider.

Test page Workaround is included
Reported by: Howard Russell.

When using <input type=image> as a submit button, only Firefox sends the name-value pair to the server.

There's an unobtrusive JavaScript fix included, which makes Opera, Firefox and IE6 behave the same (not tested in other browsers).

Test page Workaround is included
Reported by: Krijn Hoetmer.

When dynamically creating an IFRAME, its width is set correctly, according to CSS. However once the content is loaded, Explorer 6 and lower somehow forget the real width.

The workaround that I suggest is setting onreadystatechange handler to remind him the correct style.

(ppknote: resizing the page also solves the bug)

Test page Workaround is included
Reported by: Evgeny Fraimovitch.

Using css property overflow-y:scroll on iframes in IE6 will cause it to display an empty gap on the right side of the frame. The gap is approximately the size of a scrollbar. There is no workaround yet, other than refraining from using overflow-y.

Test page Workaround is not included
Reported by: Lauri Piispanen / Conmio Oy.

Some text preceded by an empty tag in minimized form (<span />), and properly styled by some CSS rules, is duplicated.

Test page Workaround is not included
Reported by: Andrea.

When using innerHTML in IE6 - aside from the tags contained in the returned string being in uppercase - an unordered list will only have an ending list item tag on the last list element.

Test page Workaround is not included
Reported by: Chris.

IE adds a white space character to the end of the text node value in list elements

Test page Workaround is included
Reported by: Alun Jones.

When multiple list items <li> with padding are floated next to each other, Internet Explorer 6 does not display them in a straight line, but instead displays them in a descending diagonal line.

Test page Workaround is included
Reported by: Ruben Arakelyan.

Let's say that you have an element with an id of id. This element can have either of two different classes, classOne or classTwo. (Either you're swapping the className in script, or it can have different classes on different pages.)

Your CSS might look like this:

#id.classOne { some styles }
#id.classTwo { some different styles }

In MSIE 6, the #id.classTwo style will never be applied - unless it is put in a separate style tag or style sheet.

Test page Workaround is included
Reported by: Isaac Z. Schlueter.

Positioning legend tags — what works and what doesn't? To be honest not a lot does.

I've tested — position: absolute, fixed, relative, floats margins and setting widths.

Test page Workaround is not included
Reported by: Marc Pacheco.

Hovering over a link causes the browser to freeze. It then becomes difficult to close down. The link should make a non-displayed element appear on hover.

Test page Workaround is included
Reported by: Chris Hester.

The event.clientX/Y values for the text portion of disabled INPUT elements is reported relative to the left/top of the INPUT element and not the viewport if the INPUT element is in a table.

Test page Workaround is not included
Reported by: Jason Brunette.

Setting certain combinations of padding, position, and white-space CSS properties on a block/inline element pair causes Internet Explorer to freeze up. (100% CPU usage.)

Test page Workaround is not included
Reported by: Martin Sutherland.

When Explorer encounters a tag with the style white-space:pre inside a table, it will only print the first line even though it renders properly in the browser.

(ppknote: IE 7 beta 2 only shows an 'O' in the print preview)

Test page Workaround is not included
Reported by: Jehiah.

If the height of a relatively positioned block is an an even number or the width an odd number, any element that’s positioned absolutly inside that block and use bottom or right will be 1px out.

Test page Workaround is not included
Reported by: Marc Pacheco.

IE crashes after updating several h1's css classes and div's display style with javascript. The bug only happens when specific styles are being used, in a specific order of updating the elements, and with specific laps of time between the updates.

(ppknote: complicated bug. Please read the page carefully)

Test page Workaround is not included
Reported by: quentin.

If you try to add style declarations in the head of a document, IE borks at the name 'style' - "unexpected call to method or property access".

I guess its getting confused between the head element <style> and the object property .style?

Test page Workaround is not included
Reported by: Ross Howard.

In Internet Explorer positioned elements generate a new stacking context, starting with a z-index value of 0. Therefore z-index doesn't work correctly.

Test page Workaround is not included
Reported by: Tino Zijdel.

If an input is given a type "button" or "submit" and the value used to apply text to the button is over 18 characters long, the button is shown stretched horizontally. The top and bottom border is also thicker. The effect increases with the length of the button text used.

Test page Workaround is not included
Reported by: Chris Hester.

When you use font-variant: small-caps, Explorer ignores text-transform: uppercase || lowercase.

Test page. Workaround is not included.
Reported by ppk.

An event handler for onmouseover an anchor should not show the URL in the status bar when you return false, and it should when you return true, but actually it does exactly the opposite.

Test page Workaround is not included
Reported by: Lal.

This bug is caused by IE not calculating position of floats correctly. As far as I can tell the left edge of the content is calculated and then the width is applied, the correct behaviour is to apply the width and then see where the content should be layed out due to any floats.

Test page Workaround is not included
Reported by: Marc Pacheco.

When you remove an element by innerHTML='' it's possible to re-append the element to the document. In Explorer Windows, though, the text goes missing.

Explorer Mac loses track of the element when it's been removed. So in this browser elements don't stay in hyperspace after they've been emptied from the innerHTML.

Test page. Workaround is not included.
Reported by ppk.

Explorer Windows and Mac, Mozilla and Opera see <fieldset>s as form fields, even though the spec doesn't mention them in the list of control types.

Mozilla and Opera allow change events on <fieldset>s, even though they don't make sense.

Test page. Workaround is not included.
Reported by ppk.

When a drop-down menu has a long option, and the form is placed inside a div of a shorter width, the option is cut off when the user scrolls far enough to the right.

Test page Workaround is not included
Reported by: Chris Hester.

Once the background-color is set on an element, subsequent 'background:inherit' rules (even those matching selectors of higher weight) can't cause the element to inherit the color from its parent.

Test page Workaround is not included
Reported by: Pierre Saslawsky.

When nesting a form wholy inside a cell certain browsers includes extra white space at the end of the form. This would appear to be a bug.

Test page Workaround is not included
Reported by: Jay Soffian.

Pages with width-defined layouts have the last several characters displayed twice when hidden inputs are included inside the layout container.

Test page Workaround is included
Reported by: Dusty.

Under some circumstances the changing the href-attribute of a link also changes the content of the link itself.

This seems to happen when the raw content of the link starts with www.something (and does not need to relatve to the real href in anyway).

Putting redundant span-tags inside the link solves this.

Test page Workaround is included
Reported by: Markus Fischer.

The INPUT (type=text/password) field in IE6 (and older probably too) has a top and bottom margin of 1px in both quirks and standard mode. This can cause a not so nice effect.

(ppknote: Opera and Safari have the same problem with the right margin)

Test page Workaround is not included
Reported by: Marcel Lipovsky.

The content box surrounding an anchor (A) element is truncated when the A elements inside a block element that inherits a non-zero letter-spacing value and appears immediately after a top-borderless DIV element.

Test page Workaround is included
Reported by: Carl Camera.

IE doesn't apply the value inline-block for the CSS display property on HTML elements that default to block level without the hack detailed here.

ppknote 4-feb-06: Workaround doesn't work in IE 7 beta 2.

Test page Workaround is included
Reported by: Spartanicus.

The following will throw an exception:

var button=document.createElement("button"); button.type="button";

The only workaround for dynamically creating <button type="button"> seems to be to convert the button HTML to a string and append it to innerHTML.

Test page Workaround is included
Reported by: Garret Wilson.

IE6 is not firing a mouseover event in the bottom padding or border areas when a DIV width is not defined, or set to width:auto.

As a workaround, setting a width brings the expected mouseover behavior back.

Test page Workaround is included
Reported by: Carl Camera.

The change event doesn't fire in IE when fields are filled from the browser's autocomplete list.

Test page Workaround is included
Reported by: Jehiah Czebotar.

When creating a table dynamically and inserting table rows with table data cells and text nodes and add it to the document tree, it is not displayed at all when you don't include a <tbody> tag. W3C DOM2 methods are used.

(ppknote: Test case does not work in Safari)

Test page Workaround is included
Reported by: Gunnar Vestergaard.

Experimenting with position: relative styles I noticed this inconsistent behavior of IE6. The image on the pages vanishes when a TD is shown/hidden, even though it is not involved in the javascript action.

Test page Workaround is included
Reported by: Chris Seeling.

In IE, when a <label> is attached to a <select> element, and the user clicks the label, the <select> is selected, but its selectedIndex becomes 0.

Test page Workaround is not included
Reported by: splintor.

If a page calls window.open(), then submitting a form on that page and returning to it using the "back" button causes the entered form values to be lost.

(ppknote: Safari has a different bug: it doesn't show the form at all when you go back. This has nothing to do with opening a new window.)

Test page Workaround is included
Reported by: Milo van der Leij.

Internet Explorer fails to render backgrounds for elements that have both an ID and a class defined. For instance, #photos.background1 will display the corresponding background in IE6 for Windows, but once that has been defined, #photos.background2, #photos.background3, etc. will not be displayed.

Test page Workaround is included
Reported by: Nathan Smith.

The document.getElementById() function returns an element with a name attribute that is equal to the id specified.

Test page Workaround is not included
Reported by: Chris Bloom.

When using a margin on a parent container with a child container inside and in there form elements then the margin of the parent container gets inherited to the form elements, and it can't be removed by a margin 0.

The affected form elements are input types, checkboxes, radiobuttons and textareas

The solution is to put an OVERFLOW: HIDDEN on the child container.

Test page Workaround is included
Reported by: Paul van Steenoven.

In any non-quirks mode doctype, content with position:relative within a block container with overflow: auto, scroll or hidden and a height shorter than its contents, will spill out past the bottom of the container.

Test page Workaround is not included
Reported by: Emmett Pickerel.

IE 5.5 and 6 set the left origin of the positioning co-ordinates to the edge of the content of the first element of the containing block, and ignore the edge of the relatively positioned containing block.

Strangely ie behaves correctly if the position is set relative to the right edge. It also behaves correctly if the containing block is positioned absolutely.

Test page Workaround is not included
Reported by: Marc Pacheco.

When applying this position: fixed workaround the following circumstances trigger a very strange behaviour of IE6:

Applying onmouseover- and onmouseout-event-handler to html-tag in combination with changing the height of the div-block via javascript to a value higher than the window will trigger the bug.

After that, hovering over the "hover-able"-area that was defined before will cause the ie6-window to shrink little by little everytime the user hovers over the displayed "onmouseover"-tag.

Test page Workaround is not included
Reported by: Tobias Gassmann.

Under certain conditions, the background image shown on a:hover remains when the mouse moves away.

Test page Workaround is included
Reported by: Ingo Chao.

In Explorer, if you need to set the accesskey attribute using setAttribute(), the attribute name passed in must be "accessKey" and NOT "accesskey".

Test page Workaround is not included
Reported by: Toh Zhiqiang.

In Internet Explorer 6.0 and higher, setting html, body {position: relative;} while in strict mode causes the browser to disallow scrolling. I don't know of any real-world applications where this would be a problem, but it's still a bug.

In Safari the entire body disappears if you set position: relative on the HTML first, and then on the BODY. Safari crashes if you check BODY, then HTML, then uncheck BODY.

Workarounds include using quirks mode, or playing it safe and avoiding setting positioning to the html element.

Test page Workaround is included
Reported by: John Hansen.

It seems that if I attempt to specify the width of a list item tag (<li>) using either a CSS class or a STYLE attribute, IE does not increment the number in front of list each item. i.e. instead of a list with items numbering 1, 2, 3, and 4, I get a 4 item list numbering 1, 1, 1, and 1.

(ppknote: This bug is very similar, but not identical, to width on ol/ul.)

Test page Workaround is not included
Reported by: Kris Oye.

Elements don't display correctly if they have a background color and a child element that uses float: left.

The #content{background: #FFFFFF;} declaration is the source of the problems. Remove it and the hidden element displays fine.

Bug applies only to Explorer 6.

Test page Workaround is not included
Reported by: Francis Mak.

When IE6 wraps an inline element with white-space:nowrap onto the next line, the element loses its left padding. Workaround is to use inline-block display instead of inline, though this doesn't work for list items.

Test page Workaround is included
Reported by: Alan Harder.

When trying to break a list of floats into rows with clear: left on the first cells in the future "rows", the cells after the cleared ones just fill up the remaining space from the previous "rows" as long as the parent's width allows.

Test page Workaround is included
Reported by: Valentin Agachi.

Setting a border around an iframe creates a padding-right and padding-bottom twice as large as the border.
This bug occurs in MSIE 6 in Strict Mode. It seems the iframe's document width is computed according to the traditional box model, while the iframe itself is computed according to the W3C box model.

Test page Workaround is not included
Reported by: G�rard Talbot..

A background on a TR doesn't work. Instead the TDs inherit the TR's background image.

Test page Workaround is not included
Reported by: Valentin Agachi.

A percentual width of an abolutely positioned element should be calculated relative to the containing block (the body, if there is no other absolutely positioned element. Unfortunately Explorer (Win and Mac) and Opera calculate the width relative to a static block that contains the HTML for the absolute element.

Test page Workaround is not included
Reported by: G�rard Talbot.

Text in a div with a background color followed by a div which floats to the left and another which is clear on the left is invisible.

Test page Workaround is included
Reported by: Justin Patrin.

A valid way to position an HTML element with the absolute type is by defining only the left offset, right offset, top offset, and bottom offset from parent. Opera and Firefox pass. IE 6 does not.

Test page Workaround is not included
Reported by: Peter Siewert.

window.opener.closed works in IE 6 but not in Mozilla or Opera due to a known bug. Also, in IE 5 window.opener.closed DOES NOT work correctly due to another known bug. I also have a longer writeup in my blog.

Test page Workaround is included
Reported by: Yakov Shafranovich.

In Internet Explorer you won't be able to view all of an absolutely positioned div because IE doesn't 'create' a scrollbar.

Test page Workaround is included
Reported by: Mark Janssen.

Using setAttribute, you should be able to set any attribute on an element. It appears that in IE, trying to set the style attribute does not work.

Test page Workaround is not included
Reported by: Dan Nye.

When using a css image, if there is no space between the end url bracket and the attributes following it, IE6 will not display the image.

Test page Workaround is included
Reported by: R.

Be careful using font-variant: small-caps in IE6, because spaces have the size of the original font and can cause positioning problems.

Test page Workaround is included
Reported by: Paul van Steenoven.

When a block set to overflow: auto contains floats Opera and Explorer Windows don't contain the floats correctly.

Test page Workaround is not included
Reported by: Marc Pacheco.

IE is wrongly case sensitive about the usemap attribute of an image map. To change the map for an image in run time, you have to set the attribute "useMap", which is against the XHTML Strict standard. In a proper situation, you should set the "usemap" attribute.

Test page Workaround is included
Reported by: Pauli Ojanper�.

In some browsers getAttribute("HREF") on an A element returns a complete URL, even if the HREF attribute in the source specified a relative path. In others it returns the exact text of the attribute in the source, which may be relative.

All browsers put the complete, resolved, URL for an A element in its .href property. In Explorer (Win and Mac) and Opera 8 the same value is returned by getAttribute("HREF"), while Mozilla (FF1.0), Safari and Konqueror (3.2.2) return the source value. I consider the latter behaviour to be correct.

The test case contains two links, the first absolute and the second relative, and reports the values of .href and getAttribute("HREF") for each. It is the last reported value that differs.

Test page Workaround is not included
Reported by: Phil Endecott.

When you try to overrule a white-space: nowrap IE5.5 ignores it.

Test page Workaround is not included
Reported by: Paul van Steenoven.

Explorer 6 Windows ignores all margins on a table. 5.0 and 5.5 ignore top and bottom margins.

Test page. Workaround is not included.
Reported by ppk.

Using a sans-serif font-family may cause an unwanted gap between tightly stacked items.

Test page. Workaround is included.
Reported by: Maxine.

The HTML specification does not allow use of a number in a class or id, it is suggested that you preceed a leading number with an underscore (_). Windows IE does not obey the rules set for a class or id beginning with an underscore.

Test page Workaround is not included
Reported by: Aaron Gustafson.

Eric Meyer's Pure CSS Popups turn out not to work in Explorer Windows unless you use one of a list of declarations to make a:hover "have layout". Meyer uses one of the correct declarations, so his example works.

Test page. Workaround is included.
Reported by ppk.
Additional tests by Claire Campbell.

It seems IE is limited in the importation of stylesheets via the @import rule. If you try to import more than 4 levels of CSS, the fifth one is not considered by Ie.

Test page Workaround is not included
Reported by: Thierry Goulet.

When creating custom DTDs like the one below, all browsers except Opera see the end of the ATTLIST as the end of the DOCTYPE. The result is that they print "]>" on the screen.

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
	"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"
[
<!ATTLIST p behavior CDATA #IMPLIED>
]>

Test page. Workaround is not included
Reported by ppk.

If you give TDs a CSS width and then use colspan, Explorer Windows still uses the declared CSS width for the content of the colspanned TD.

Test page. Workaround is not included.
Reported by ppk.

For nested boxes IE doesn't add the outer box padding and the inner box margin, as all other browsers do.

Test page Workaround is not included
Reported by: Daniel Bleisch.

The CSS 'bottom' property can be used to place an absolutely positioned element relative to the bottom of the containing block. When 'top' is also specified, the element should stretch from 'top' to 'bottom'. However, when both are specified, IE ignores 'bottom' and instead bases the height of the element on its contents (shrink-to-fit).

IE similarly ignores the CSS 'right' property when 'left' is specified.

Only IE 6.0 and Firefox 1.0 tested. No pure CSS (non-javascript) workaround found yet.

Test page Workaround is not included
Reported by: Justin Schiller.

In Explorer Windows an element both floated and cleared is ignored by a subsequent float: the latter is positioned higher than the former, and overlaps it.

Explorer Mac and Opera don't get it quite right, either, but they're not as bad as Explorer Windows.

Test page Workaround is not included
Reported by: Bruno Fassino.

When the font of a page in an iframe is resized, Explorer Windows seems to retain the line-height of the previous font size.

Test page. Workaround is not included.
Reported by ppk.

IE/Win does not respect a set line-height in presence of an inline image shorter than the line-height.

Test page Workaround is not included
Reported by: Bruno Fassino.

When navigating between named anchors in IE, the window.location object changes to reflect the change in the URL. This is correct. However, the window.location object should also change when you use the back or forward buttons, or the history.back() and history.forward() methods, to move between these anchors, and in IE 6.0 and IE 6.1, it does not.

Opera has exactly the opposite bug: it never shows the hash.

Test page Workaround is not included
Reported by: Isaac Z. Schlueter.

Internet Explorer on Windows completely ignores closing LI tags.

This means that if the list item contains a block level link, the blank space after the LI will be rendered as a blank line.

It also means that if elements are nested within the list, but not within a LI (this is invalid HTML, but people still do it), IE will treat them as if they are in the preceding LI. This modifies both the DOM and the response to CSS.

Test page Workaround is included
Reported by: TarquinWJ.

The box model of a td turns out to be unchangeable in Explorer (Win and Mac), Mozilla and Opera. Explorer Mac forces it into the traditional model, while the other browsers force it into the W3C box model.

Therefore you can't switch the box model of a td, something that is possible for any other element.

Explorer Windows, though, switches the TDs to the correct box model when you use table-layout: fixed. Unfortunately this does not work in the other browsers.

Test page. Workaround is included only for Explorer Windows.
Reported by ppk.

Using the :first-letter selector (such as to do drop-caps) crashes IE 5.5 and 6.0 WinXP when that first letter is part of an underlined link which is visible on-screen.

Reported to MS in February 2004 - may be fixed in SP2, but I haven't been able to verify it.
:first-letter doesn't work in IE 5.0 Windows, so that browser doesn't crash.

Rare in that this crashes the browser, rather than just displaying in a funny manner.

Test page Workaround is not included
Reported by: Dave Polaschek.

Explorer 5.5 Windows and below see an imported print style sheet as a normal screen style sheet.

Test page. Workaround is not included.
Reported by ppk.

When placing an absolutely positioned element in a relatively positioned one, Explorer 5.0 Windows may calculate the left relative to the browser window instead of to the relatively positioned element.

Test page. Workaround is included.
Reported by ppk.

Explorer positions a background-image underneath the border of the element. Therefore it's impossible to get the position of a background image in an element with a border correct in all browsers.

Test page. Workaround is not included.
Reported by ppk.

When using multiple class names for an element, Explorer will ignore all but the last of class names:
element.class1.class2 { }
Explorer completely ignores the class1 selector and happily applies the rule to an element that has only class2 set.

Test page Workaround is not included
Reported by: Tino Zijdel.

In IE with SP2 if you a few a few anchor tags for a menu that have some vertical distance from each other (by using the bottom margin) and on hover want them to change background color, you are in for a nice surprise (removing the margin of the anchor 2 places behind it ..)

Test page Workaround is not included
Reported by: Petrioli Gabriel.

Setting an inline border style (element.style.border) works in all browsers. Removing it to allow the normal border styles to return, however, is tricky in Explorer Windows and impossible in Safari.

Test page. Workaround is included.
Reported by ppk.

Setting up a row of floats in Explorer 6 can lead to a plethora of erratic behavior, namely duplicate characters; in this bug report, I go into detail on some pure-CSS triggers which can make these extra characters appear, examples on how the bug is triggered, and different ways to combat it.
Note that these bugs are not caused by comments in your XHTML, as is the well known original Duplicate Characters Bug.

Test page Workaround is included
Reported by: Agent EQzE.

When a floated box does not have non-floated content alongside it that extends the full height of the float (including its vertical margin), the next 'cleared' div after the float has a top padding that is twice what it ought to be (in IE5.5 and 6).

Test page Workaround is included
Reported by: Luke Plant.

When you apply a width to an ol or ul, Explorer Windows secretly applies more rules. Moreover, 5.0, 5.5 and 6.0 each apply different rules.

(note: This bug is very similar, but not identical, to Ordered list rendering with width on LIs.)

Test page. Workaround is not included.
Reported by ppk.

IE applies HTML normalization to the data that is assigned to the innerHTML property. This causes incorrect display of whitespace in elements that ought to preserve formatting, such as <pre> and <textarea>

Test page. Workaround is not included.
Reported by Sebastian Redl.

When you define visibility: hidden on * Explorer Windows and Safari don't show anything ever again.

Test page. Workaround not included.
Reported by Giuseppe Bertone.

font-weight: 600 and font-weight: bold are not equivalent. The browsers stretch up their fonts when you use 600. For a really correct bold font you must use bold or 700.

Mozilla exhibits a similar bug only on Mac: any text with font-weight: 600 appears as normal, non-bold text.

Test page. Workaround is included.
Reported by ppk.

When you change the font size of button elements, odd things start to happen in IE Win & Mac, and Mozilla. Of course, each browser has its own take on 'odd things'. The Mozilla problem is solvable.

Test page. Workaround is included.
Reported by ppk.

Up to XP SP2 Explorer 6 Windows supported @i url(some.css).

Test page. Workaround is not included.
Reported by ppk.

Any button uses the traditional box model instead of W3C's, regardless of rendering mode.

Test page. Workaround is included.
Reported by ppk.

If you move a checkbox through the document by W3C DOM methods, the checkbox returns to its default state (checked or unchecked).

Test page. Workaround is included.
Reported by ppk.

If an iframe contains body {position: relative}, the scrollbars of the main window may disappear.

Test page. Workaround is not included.
Reported by Keith Swartz.