Steve, thanks a lot! I never thought the problem is in avwmerge. The Analyze data were converted from another format through a idl program i wrote. I had the output filenames zero padded when i did the human subject data, but ignored when i did the simulated data. > Oh - sorry, forgot to say - your parametric activation in the data > (time-varying height) doesn't fit the model you're using!! yes, the EV file we provided is an unweighted one to see the mean effect. In experiment we applied 5 forces of different level on the subject in one run, and each force applied only once. We are more interested in seeing the linear effect from using 5 EVs in the setup. Thanks again for spending time working on it! Heather