I think that object orientation with polymorphism is a great way to encapsulate complexity and decouple it, from the standpoint clarity, reusability and reliability. But the ability to define 01 level record formats in COBOL is still a desirable feature, for occasional use. Every now and then it would be useful to have a general-purpose struct that you can redefine easily. Compiler/interpreter bugs: I don't know what operating system you are referring to, but Unisys COBOL was perfectly solid. Repeated blocks of code that do exactly the same thing: It was possible to write COBOL that way, but we were taught structured programming, which is as modular as VB6. Add OO to COBOL and you do the same thing for COBOL that .NET did for VB6. Loop through a thousand records to find a single customer: You seem to be referring to reading a tape or a flat file. Unisys had an excellent database, called DMSII, which could do many of the things you can do in a modern relational database, although it did not have SQL. The experts tell me that DMSII was particularly strong in its ability to do a synchronized recovery. We were strongly discouraged from writing linear searches in DMSII. In short, I am grateful to be an OO programmer today, but I respect COBOL. ageless