I'm not sure I under stand the last part of your question. But regardless, You can simplify your processing by realizing that you have a delimited input file. Records are separated by // flags are separated by /. So with judicious use of split you have:

#!/usr/local/bin/perl

use strict;

# pull everything into a string
my $str = join("",<DATA>);

# dump the newlines
$str =~ s/\n/ /g;

# loop through the records (// delimited)
foreach (split(m!//!,$str)) {
  last unless /\S/;                       # skip that pesky last blank
+ record
  my ($title,@flags) = split(m!/!);       # break out the fields (/ de
+limited)
  $title =~ s/\s+/ /g;                    # kill extra whitespace
  print "References:$title\n";            # print the title
  foreach (@flags) {                      # loop through the fields
    my ($key,$value) = split(/=/);        # split into pairs
    $value =~ s/\s+/ /g;                  # kill extra whitespace
    print ucfirst($key)."\t$value\n";     # print each one
  }
  print "\n";
}

__DATA__
TITLE     An excitatory scorpion toxin with a distinctive feature: an
            additional alpha helix at the C terminus and its implicati
+ons
            for interaction with insect sodium channels
             /interaction_site="Q8, N9, Y10, N11, C12, F17, W38, R58, 
+V59 and K62 form the putative bioactive
                surface in mature toxin (Zilberberg et al., 1997)."
             /channel="Sodium channel"
             /target_cell="Insect specific (Excitatory)"
             /c_end="Free"
//
TITLE     Cloning and Sequencing of an Excitatory Insect-Selective Neu
+rotoxin
            BmKIT cDNA from Buthus martensii Karsch
            /interaction_site="Sequential deletions of C-terminal resi
+dues suggested Ile73 and Ile74 for toxicity. {Oren et al., 1999}"
            /channel="Sodium channel"
            /c_end="Free"
//
[download]

This approach works well if the files are small. The initial read the whole file and stuff it into a string breaks down if the file is huge. If the input files are huge it can be modified to read one line at a time until it has read a full record, then stuff the record into a string and parse it. Like so:

#!/usr/local/bin/perl

use strict;

my $str; # holds the records

#loop through the data
while (<DATA>) { 
  chomp;                                    # kill newlines
  if (m!//!) {                              # we have a record
    my ($title,@flags) = split(m!/!,$str);  # break out the fields(/ d
+elimited)
    $title =~ s/\s+/ /g;                    # kill extra whitespace
    print "References:$title\n";            # print the title
    foreach (@flags) {                      # loop through the fields
      my ($key,$value) = split(/=/);        # split into pairs
      $value =~ s/\s+/ /g;                  # kill extra whitespace
      print ucfirst($key)."\t$value\n";     # print each one
    }
    print "\n";
    $str = "";                              # zero the input buffer
  } else {
    $str .= " " . $_;                       # accumulate data
  }
}

__DATA__
TITLE     An excitatory scorpion toxin with a distinctive feature: an
            additional alpha helix at the C terminus and its implicati
+ons
            for interaction with insect sodium channels
             /interaction_site="Q8, N9, Y10, N11, C12, F17, W38, R58, 
+V59 and K62 form the putative bioactive
                surface in mature toxin (Zilberberg et al., 1997)."
             /channel="Sodium channel"
             /target_cell="Insect specific (Excitatory)"
             /c_end="Free"
//
TITLE     Cloning and Sequencing of an Excitatory Insect-Selective Neu
+rotoxin
            BmKIT cDNA from Buthus martensii Karsch
            /interaction_site="Sequential deletions of C-terminal resi
+dues suggested Ile73 and Ile74 for toxicity. {Oren et al., 1999}"
            /channel="Sodium channel"
            /c_end="Free"
//
[download]

This should also print the tags regardless of the number of lines. hope it helps.

-pete
Entropy is not what is used to be.