Overall, FileSeek looks like a pretty slick tool. However, I do have two performance related questions/suggestions. My D: disk has 40,000 subdirectories. It takes FileSeek ~1minute (with an empty query) to find all files "foo.*". However, the Windows command 'DIR /S foo.*" only takes 5 seconds. The CPU utilization for FileSeek was 8% on a 12 CPU system (i.e. 100% on one CPU). Perhaps this is due to updating the "Current" text box? If so, perhaps updating the progress could be made an option, or done less frequently (say, 4 times a second). Also, I noticed that accessing the disk seems a little slow (3MB per second rather than the 70MB that the disk is capable of). This seems to be a semaphore related issue, as the CPU is not being maxed out either. If this code is written in Microsoft C or C++, have you considered using the *_nolock() versions of getc() etc. found in the Microsoft stdio.h file? This can increase your disk I/O performance by 4x or more.