There is a difference. For example, DBI's do() returns the number of rows affected -- but some statements succeed even though no rows were affected. This allows for the following bit of code to be sane:

my $rows = $dbh->do(
   'UPDATE MyTable SET note_date = GETDATE() 
    WHERE ErrorCode IS NOT NULL'
);

if ($rows) {
   # true, even if zero rows were affected.
   print 'Found and updated rows: (',$rows,")\n";
}
else {
   warn 'Error during update query';
}
[download]

Having a query affect zero rows is perfectly valid, and doesn't necessarily mean an error. This makes even more sense when other methods in the same program return '0' as the false value rather than undef: it allows for programmer choice and a consistent read.