Hi Doug,

Thanks for your reply. Yes, it does work with a longer -refeventdur. In 
fact, I had tried that yesterday before posting my original message but 
I'd forgotten to make the corresponding changes to the 3rd column of the 
paradigm file and so kept getting the error message.

The shortest I can bring -refeventdur down to is 0.05. Anything smaller 
causes the same error as before. In the end it's not a huge problem for 
us if we have to set an event duration of 50ms instead of 16.6 ms but 
I'd like to understand what is going on.

So, I wondered if the error had anything to do with the fact that when I 
run mkcontrast-sess the script automatically sets -TER to 0.05. I re-ran 
the analysis and set -TER and -refeventdur to 0.016 in mkanalysis-sess. 
The analysis runs with no
 error for selxavg-sess and visually (with 
tksurfer) the results look the same as with -TER and -refeventdur set to 
0.05.

Since the help for mkanalysis-sess -TER says " It is best to know what 
you are doing if you are going to change this.",  I'd like to know if it 
makes sense to change the TER like I have. What does the TER achieve (if 
that's not too complicated a question)?

Thanks again,
Leila