Don't add on error resume next!

I see you use on error resume next at many spots, you shouldn't need this if the code is right, check your code and make sure you don't need to use it; it'll slow down the program in whole.

Also try using signature scanning within files instead of loading files as a library, using the signatures you can more quickly find the areas resources are located you want to extract (unless a file is compressed) and it'll speed up the program; most formats allow a way to determine the size of the resource so you know how many bytes to extract from the found offset!

I see SWF and compressed SWF support is already within the program, why isn't enabled?

By not using API calls for extraction as a base you won't require administrator privileges on Vista and Windows 7, however, you can add a resource to your program to force administrator privileges, most end-users don't even know how to run as administrator!

Apart from this all, the code might be speeded up easily by changing some unicode commands to single byte commands; furthermore I'd use a resource for the in-program texts instead of setting them within the code...