You could always gather up the data into a list of array refs, each pointing at the block data elements. If the raw data file is small enough, you could always slurp the whole thing, and process it as below.

#!/usr/bin/perl

use strict;
use warnings;

undef $/;

my @blocks = map { [ split /\n/ ] }
    ( split /(?=ITEM)/, ( map { s/#.*\n//; $_ } <DATA> )[0] );

print "$_\n" for @{ $blocks[1] };    # print block no. 2

__DATA__
#This is the ITEM file
ITEM NO:1
[aaa]
111
[bbb]
222
[ccc]
333
ITEM NO:2
[ddd]
444
[eee]
555
[fff]
666
ITEM NO:3
[download]

Yeah, it's kinda golfy.

Hope this helped,
-v.