Here is what I cam up with as a solution using split. I was looking at the Perl Cookbook and the ideas here. This seems to work but is a bit slow with a million records.

#!/usr/bin/perl
use strict;
my @chunks;
{

local $/ = undef;
@chunks = split(/^(?=(?:EN))/m, <>);

}
print "Rec: ",scalar(@chunks), " eor\n";

for (my $i=0; $i <= scalar(@chunks); $i++) {
        $chunks[$i] =~ s/[\r\n]+//g;
        if ($chunks[$i] =~ /^EN/ ) {
                print "$chunks[$i]\n";
}
}
[download]

The bottom bit was just to prove it worked and print out new clean records. I need to do some addtional error checking but this produces a clean file.

Any ideas for improvement?

sxmwb