You know what they say… it's not what you don't know that will hurt you, it's what you know that isn't so! In other words, your misconceptions. Or, as Paul Nielson calls them in his SQL Server Bible… MYTHconceptions.
Some misconceptions come from misunderstanding of complex information, or from misinterpreting your own results, and assuming we can generalize behavior from one particular situation. Since I teach advanced classes to students with lots of SQL Server experience, I actually see a lot of misconceptions around things that used to be true, but aren't true anymore. I would imagine it's next to impossible to retest every single thing you know about SQL Server, every time a new version comes out. One simple example would be if someone believes that bit columns cannot be nullable. That used to be true, but not now. Another current myth I hear occasionally in my classes is that a TRUNCATE TABLE operation is not logged and cannot be rolled back. This was true a long long time ago, but hasn't been true for many versions now.
Paul Randal is doing a wonderful job of pointing out some of the most prevalent and/or dangerous myths, providing details about a myth a day for 30 days, and has gotten through 8 days so far… he starts his series here.
I love to hear more examples of things that you thought were true (perhaps because they used to be) but aren't anymore!