Huzzah, the crufty re-implementation of Bio::SeqIO continues unabashed by suggestions to the contrary and pointers to its documentation.

Addendum: That perhaps sounds a bit cruel, but if you don't understand enough Perl to know how to split scalars into characters you really shouldn't be trying to re-invent wheels that other people have already solved. Use the off-the-shelf widgets for the parts you can, and get fancy once you know what you're doing. Something like Beginning Perl for Bioinformatics that's aimed at biology-types specifically should help you get going.

Perhaps I am not understanding you correctly, but this seems to be a very different question than the one you asked first. In the example you gave, what happens to the values in $entry, $title, $org or $acc? You seem to drop them on the floor ...

To solve this kind of a problem, I would probably establish a hash of array references, something like this (untested):

use strict;
my %filecontents = ('entries' => [],
  'titles' => [],
  'organisms' => [],
  'accessions' => [],
  'unknown' => [] );

my $filehandle;
unless (open($filehandle,"<file.txt")) {
  die "Cannot open file.txt: $!";
}
while (<$filehandle>) {
  chomp;
  if (/^ENTRY/) {
    push @{$filecontents{'entries'}}, $_;
  }
  elsif (/^TITLE/) {
    push @{$filecontents{'titles'}}, $_;
  }
# and so on ...
}
[download]

Update: Notice, by the way, that I check for success in opening the filehandle, always a good thing to do.

Update 2: Arrghh! I was duped! I didn't realize you'd asked this question before. Why not pay attention to the answers you've already received, as Fletch points out below?