First, look at the perlmonks Site How To page for instructions on using the <code> tag when posting your code -- it will retain the indentation and make sure that things like square brackets and angle brackets come through as intended.

Second, consider a different approach. From the data samples you provided, it appears that there is no "key" field that serves to cross-reference the "parent" file with the "child" file. That means you need to be confident about how these two files were created: for each line in the parent file with "C" in the fifth field, exactly one line is written to the "child" file. If this is accurate, then a workable approach could go like this:

• Open the "child" file, read all its contents into an array, and close that file.
• Now, open the "parent" file; for each record that contains a "C" in the fifth field, use perl's "shift" function to take the next entry off the array of "child" records, and append this to the current "parent" record.

You shouldn't have to worry about memory constraints unless the "child" file has hundreds of megabytes of content. The code could end up looking like this:

use strict;

open( IN, "ICD_MEAS_child.txt" ) or die "failed to open child data: $!
+\n";
my @children = <IN>;  # read them all at once;
close IN;

open( IN, "ICD_MEAS.txt" ) or die "failed to open parent data: $!\n";
while (<IN>) {
    if ( /,C,/ ) {
        my $child = shift @children;
        s/\n/,$child/;  # update: added the essential comma
        s/,/;/g;  # update: added this line, as per orig. post
    }
    print;
}
close IN;
[download]