You might get success from the glm function in R. If the response was coded as binary with 1=success and 0 = failure and define a factor 'group' with two levels and co-variate 'x' then a call of form:

glm(response~group+x,family=binomial())

would fit the model you describe and should replicate the SAS results.

Thank you Stephen. I think I didn't explain enough.

Here is a new question: What is the difference, in SAS, between the strata statment and adding the factor in the independant variables? In fact, I did not found the same results with those two methods. I don't need to estimate the parameter of the stratification variable, just being conditional on it.

I agree that this is quite the same results, and maybe it will be enough.

Sorry Audrey, I've got no expertise in SAS. I thought you were asking how to fit that model in R. Good luck

PROC LOGISTIC in SAS with STRATA statement performs the conditional logistic regression. The same result you obtain in R using clogit and specifying strata. clogit is inside library survival.

If you want separate logistic models for a variable, I think iwith SAS you would use the BY statement. I have never used it with PROC LOGISTIC, but you can try.

Yes Stephen, I would like to do the same in R...

Thank you Francesca, but with the same database and the same model fitted using SAS with STRATA statement and using R with clogit{survival}, SAS converged, R did not. I will continue to search.

Starting from the same stratified database I obtained same results

You could also have a Random Intercept Effect in you model.