The only known problem with GLMStat is the calculation of degrees of freedom for certain models (eg Poisson models with accidental zeros), a problem shared with many other programmes (eg GLIM 3.77). See McCullagh and Nelder 2nd edition p207-208 for an example.
In the event that you find a bug in GLMStat please either write to me at the address given in 'Ordering Information' or during 1994 you may email me via 'cgacb@cc.newcastle.edu.au'. As this is not my account it may take some time to reach me. Please try to include as much information as possible eg data used.