It's possible that some files were heavily fragmented and for whatever reason recuva couldn't quite put the right pieces together. I'm just theorizing on this next point on how Recuva might work, but one of many possible methods of data recovery is to locate and identify a table of contents (addresses of files on a disk) which might be "current," or might be an old version (from before some files were moved or otherwise rewritten), or might be partially corrupt, and then use this table to locate files. As a simplified example, a file might have an address of 12345, which in binary is 11000000111001. Through some random corruption, perhaps a hardware error or exposure to a magnetic fields, one random bit got changed; perhaps the first 1 changed to a 0, resulting in 01000000111001, which is 4153 in decimal. A recovery program might read this incorrect address of 4153 and then look in completely the wrong area of the disk for the file's data. I had a 2TB drive recovered (using some other software), and I would guess that 5-10% of the recovered files contained the wrong data. I was even able to recognize the some of the incorrect contents, e.g. one recovered image file contained a chunk of text that come from web pages, so it wasn't just random binary garbage (although image data replaced with text does amount to garbage output).