OK, I think I know what you doing. Why not store the RECNO() in the cursor and not the Memo contents, and then when the condition ? has been met just do a Seek on RECNO and retrieve the Memo contents of that record and then do whatever you where doing to it.

After reading the other new posts you will have to detail exactly what you are trying to achieve for us to be able to answer your question, manipulating 100MB in a memo is going to required some very specialized processing.

>Hmmmm. I see what you are saying. This does bring up one possibility, but my observations are kind of contrary to what you just stated.
>
>I can 100% say this: the size of the replacement string NEVER exceeds the size of the string in the memo in the first place. With that being said, it kind of defies your suggestion, as I am sure the string will never be replaced with something larger.
>
>By the way, I don't even have to scan, this cursor ALWAYS has one record. Over the years it has evolved that this cursor only represents the current record being "processed", and thus, is 1.
>
>I'm going to try something in the "parent" process. The first time the routine is called I am going to replace the field with the field plus 20% then back to the original. I will then see if there is any bloat...
>
>Testing...
>
>Done. Memo bloat still happening. Argh.
>
>To summarize:
>
>Before any processing of a record occurs, I replaced the contents of the memo field with the memo field plus space equal to 20% of the memo field. I then replaced it back with the original field. Processing then began and the replacement string was never bigger than the original value. Memo bloat occurred.

Regards N Mc Donald