Hi, I have a data set with values (L) varying from 1 to 700, with repeated numbers, and I want to sample them according to the probabilities given by a logistic curve P=1/(1+e(-r*(L-Lc))) where r gives "S" the inclination and Lc the "rotation point". P ranges from 0 to 1. As I could see I cannot use this P as prob in sample function. I'd need the probabilities as given by dnorm function, isn't it? Well, how can I ?translate? the P given in the logistic function? Thanks for any help. Best regards. Antonio Olinto Sao Paulo Fisheries Institute Brazil ------------------------------------------------- WebMail Bignet - O seu provedor do litoral www.bignet.com.br