As I understand, when mle is given derivatives of the log-likelihood, it checks the computed derivatives against its own numerical derivatives at the first iteration. If they are different, the program reports that there is a difference and stops. This is a very useful test, but the diagnostic information could be a bit more helpful. Would there be any problem in reporting the computed derivatives and the numerical derivatives, because this would help the programmer identify where the error in programming the derivatives has occurred? At the moment I am struggling with an unusually messy set of derivatives. While I can, I think, persuade the program to generate the comparison, it would be much easier if this were done automatically before the program stops. As a further thought: what is the test that the numerical and computed derivatives are different? I can imagine circumstances where one might start the estimation from a potentially degenerate set of initial parameter values that might lead to a false positive for the difference between the two. Gordon