Dear R-users, How can I impose limits on parameter estimates in NLME as indicated in the second #'ed line below? Can I use the naPattern, some thing like this: nlme(response = yftp(x,ii,cc,b,dd),... , naPatternis.na(response) ) ? yftp<-function(x,ii,cc,b,d){ tpx<-tpx(ii,cc,b) # x value for top point of 2nd order polonomial tpy<-tpy(ii,cc,b) # if (ii<0 || ii>1 || cc<0|| cc>1|| b<0) {print(NA); return(NA)} return( (ii+b*x)*(cc-x)*(x<=tpx) + tpy*(x>tpx)+ d*x) } Best whishes, Karsten