An exit status of 0 means no differences were found, 1 means some differences were found, and 2 means trouble.

But, the situation is a little different from within a Perl script. From system

The return value is the exit status of the program as returned by the wait call. To get the actual exit value, shift right by eight (see below).

print $? >> 8, "\n";
[download]

Update: major re-wording.

Try running the command from the command line and displaying what windows sees as the exit code before perl gets it hands on it:

c:\test\xx>diff junk5.png junk6.png >nul & echo %errorlevel%
0

c:\test\xx>diff junk5.png junk4.png >nul & echo %errorlevel%
1
[download]

The likely cause is that whilst Windows exit codes are ints, Perl internally truncates them to 8-bits before concatenating it with other bits that are meant to emulate child termination reasoning per the system docs:

If you'd like to manually inspect system's failure, you can check all 
+possible failure modes by inspecting $? like this:
    if ($? == -1) {
        print "failed to execute: $!\n";
    }
    elsif ($? & 127) {
        printf "child died with signal %d, %s coredump\n",
            ($? & 127),  ($? & 128) ? 'with' : 'without';
    }
    else {
        printf "child exited with value %d\n", $? >> 8;
    }
[download]

My win32 system does not have diff as a built in command. My guess is that you downloaded it. It would be difficult to help you without knowing what program you are using.

From memory and experience often 'diff' type programs return 'succesful' when the compared files are identical and return '!successful' when the files differ. The typical thought is that it is more useful to use the exit code to indicate file difference (or lack thereof) in addition to if the command ran without error.

Typically the exit code will be one value (say 1) if the files differ, another value (say 0) if they are the same, and a third value (say 2) if there was an error.

Hope this helps.

Update: typo found and fixed.

$ echo Hello >t1.txt

$ echo World >t2.txt

$ perl -wle " $ret = system q[diff -u t1.txt t2.txt >NUL 2>&1] ; print
+ $ret; print $?"
256
256

$ perl -Mautodie=system -wle " $ret = system q[diff -u t1.txt t2.txt >
+NUL 2>&1] ; print $ret; print $?"
"diff -u t1.txt t2.txt >NUL 2>&1" unexpectedly returned exit value 1 a
+t (eval 10) line 13
 at -e line 1

$ perl -MIPC::System::Simple=system -wle " $ret = system q[diff -u t1.
+txt t2.txt >NUL 2>&1] ; print $ret; print $?"
"diff -u t1.txt t2.txt >NUL 2>&1" unexpectedly returned exit value 1 a
+t -e line 1
[download]

$ diff --help |grep -i exit
Exit status is 0 if inputs are the same, 1 if different, 2 if trouble.
[download]

$ perl -MIPC::System::Simple=system -wle " $ret = system [0,1], q[diff
+ -u t1.txt t2.txt >NUL 2>&1] ; print $ret; print $?"
1
256

$ perl -Mautodie=system -wle " $ret = system [0,1], q[diff -u t1.txt t
+2.txt >NUL 2>&1] ; print $ret; print $?"
1
256
[download]

Yeah - thanks guys.
Like you say, it's just diff doing what it's supposed to do.

Cheers,
Rob