Yesterday, because the action that failed so spectacularly was non-essential to me, I enjoyed Testopia's combination (pictured above; click for full size) of Z-axis violation, inappropriately large and inappropriate content wedged into a dialog, unexpected text box in the same dialog, likely lack of validation of the server's response, and the tasteful use of colour and font size to reinforce the nature of an internal error while at the same time obscuring potentially useful debug information by truncating it.
This one is glaringly obvious, but it's as well to remember from time to time that the tools we use to test are themselves applications with their own bugs. How well have the external tools you use been tested? How extensive is the coverage of that supposedly standards-compliant open-source library you've just incorporated? What effects could this have on your work? How irritating is it when you end up investigating serious-looking issues that lead back to bugs in a third party tool? Can you use this as motivation to make your product as good as it can be for your users?