Todd: Most likely it's a memory issue. Does SPM write out ResMS images? If not then it is failing at the estimation of temporal non-sphericity. In any case you might want to put in a breakpoint using debug at around line 745 of spm_spm and then step through the rest of the estimation to see where the problem is. By doing this you could also save variables manually along the way. Sometimes starting matlab without java will hepl with calculations that use a large amount of memory. However, you cannot use the edit facility without java. Darren At 02:45 PM 4/13/2005, Todd Thompson wrote: Hi, all. I have a hopefully simple novice question: We're running a GLM for an fMRI experiment with ~600 regressors. The GLM takes around four hours to run, and outputs all our beta.img files. Unfortunately, it will often crash when writing the SPM.mat file. Does anyone know what causes this, and how to fix it? There's no matlab crash dump, though the matlab process abruptly terminates. Thanks! Todd ---- Todd Thompson Helen Wills Neuroscience Institute University of California, Berkeley ------------------------------------------------------------------------- Darren R. Gitelman, M.D. Cognitive Neurology and Alzheimer¹s Disease Center Northwestern Univ., 320 E. Superior St., Searle 11-470, Chicago, IL 60611 Voice: (312) 908-9023 Fax: (312) 908-8789 -------------------------------------------------------------------------