It's funny how nobody has been able to give a good answer yet. I'm not sure this one is either. You should probably take it as a hint that they should be used as little as possible. They're after all procedural rather than OOP.

Here are some more examples:

In Obj-C, where they're called class methods, they're commonly used as allocation wrappers where the object is put in the reference counting pool before being returned.

Another example from Obj-C is to register a new class in to a class collection. Let's say you have a set of classes each handling one type of file. When you create a new class for a new file type you could register it to the collection (a global variable) using a static method in the class that determines the type of file.

In C++ another use I can think of is to softly catch errors. Your constructor function can't fail, except by throwing an exception. You could set an error instance variable, but that's not always appropriate. Instead you can do the parts that might fail in a static wrapper and then allocate and return the new object, or NULL on failure.