eg- The reason I was messing with the input record seperator was because the text in the fields of the file seemed to contain the default record seperator. If I loaded the data file in Xemacs, it would show the problem text fields containing ^M where as the records themselves were seperated with a return. Both the record seperator and the problem text areas showed as "0D 0A" in a hex editor, so I guess they are identical. I am sure I am missing a simple thing here, but using $/ as "|\n" (every record finishes with a |) is working. Thanks for the reply. I hope I have not made too much of an idiot of myself on my first post....