<<snipped>> I would expect to hear squeals of protest from the software developers. We would depend on the reviewers and journalists to do or commission these relatively simple tests in any report they made on the various genealogy programs. <<snipped>> Tim - actually, I doubt that any serious software developer, by which I mean the actual coders themselves, would have the slightest objection to your idea, as it's a fairly standard idea, known as round-tripping. The squeals are more likely to come from the guys running the companies who would want to redefine the question to the benefit of their software. Mentioning no names but one company would protest that interfacing to online trees is more important. In truth, round-tripping is only a component of the test suite as it's possible that a misunderstanding cancels out - for instance, I might have misunderstood which was baptism and which was christening. Such a mistake is likely to be reflected in both my import and my export, thus getting us back in the same place. Which is why your 2nd set of tests (use the software to create the data) is also important. The issue is that, while your tests have a simple principle, there's a lot of data there to be compiled. While there are bloggers who have GEDCOM files of an evil size to load as tests, I doubt there is any guarantee of a percentage coverage of the GEDCOM standard. In addition, the test data needs to be compiled in such a way so as to hit all possible combinations - you can't just tick every feature off, because if you load items 1, 2 and 3 OK, that's no guarantee that 1 followed by 3 won't cause the import to crash. So - all do-able - but time consuming. Adrian B