G'day mao9856,

Here's another (less busy) way to do it.

#!/usr/bin/env perl

use strict;
use warnings;

use Inline::Files;

my %match;

++$match{substr $_, 0, 3} while <MATCH_DATA>;

while (<PARSE_DATA>) {
    print if $match{substr +(split)[1], 0, 3};
}

__MATCH_DATA__
ABCD12
XYZ13
EFGT45
UVWZ34
TSR78
__PARSE_DATA__
ID121 ABC14
ID122 EFG87
ID145 XYZ43
ID157 TSR11
ID181 ABC31
ID962 YTS27
ID529 EFG56
ID684 TSR07
ID921 BAMD80
[download]

Output:

ID121 ABC14
ID122 EFG87
ID145 XYZ43
ID157 TSR11
ID181 ABC31
ID529 EFG56
ID684 TSR07
[download]

I've used Inline::Files just to show the technique. It's good you've used the 3-argument form of open; but less good that you've used package variables for the filehandles — prefer lexical filehandles instead. Also, your error reporting (i.e. or die) is rubbish: either spend a lot more time on this tedious and error-prone task yourself, or just let Perl do it for you with the autodie pragma.

Please post data within <code>...</code> tags as you did with your code. This makes it a lot less work for you; your data isn't subject to HTML interpretation (e.g. special characters and whitespace compression); and it makes it a lot easier for us to paste it directly into any example code we might provide.