Can you get a confidence interval of an independent variable from directly out of PROC LOGISTIC? I'm using LOGISTIC to model prob (survival) as a function of a continuous parameter (TNF50) I'm using a basic model statement SURV = TNF50 I need to estimate TNF50 with 95%CIs for various specified survival probabilities I can easily use the parameter estimates produced to work backward to TNF50 but how do I get the 95%CIs for TNF50? I believe you can do this using PROC PROBIT with the INVERSECL option but I can't see this in LOGISTIC Any thoughts anyone? Thanks Alison