Common usage is : du -sh file(s) name(s) or du -sh /path/to/dir/*
Replace 'h' by 'k','m' or 'g' for Kilobytes, Megabytes and Gigabytes instead of human-readable. With k/m/j switches, you can even pipe the output un sort -n (numeric sort instead of lexicographic) to get the sorted by size list of files in a directory.
If you still have a big difference, you may want to try the --apparent-size switch to du which will allow you to diagnose sparse files. (files with empty space inside, to be simple)
