Basically, if the file you are downloading is large, there's a greater-than-zero chance that part of the download is corrupted (essentially, if during the period of download, you receive more that 2^32 packets and there is some agent in the connection that tends to cause bit errors, chances are you will have a checksum collision, or an undetected corruption).
Some sources for larger downloads (e.g. Linux DVD ISOs) will provide MD5 and/or SHA(128|256) "hashes" for the file you are downloading. Essentially, once you've completed the download, you use a tool (like this
for either type and many more). Run the tool on your download and ensure that the resulting number (it will be a hexidecimal number) matches what is indicated by the download source.
Hashes have the property that a slight difference between two file will result in (with a high level of probability) wildly-different hashes, so it should be pretty easy to spot differences if there was a corruption in downloading. As an example:
brad@onyx:~$ echo "this is a test" > test1
brad@onyx:~$ echo "this is a tess" > test2
brad@onyx:~$ sha256sum test1
brad@onyx:~$ sha256sum test2
If your download source does not
provide a hash for comparison, you could simply attempt to re-download it and compare the hashes of the two files