Allan and Ignacio, thanks for the replies. I am still unable to get the expected number of observations for the btilda variable when I restrict the sample within a Monte Carlo loop. (I was already opening the new data file (test.gdt) but, as I said, it only gives me back one observation back for btilda, even if I do the loop --progressive). (When I do it without restricting the sample it works perfectly). I am working with version 1.4.1 for Windows XP. What could be wrong ? Thanks again, Maite