I have various samples on which three survival time variates have been collected. How can I compare the difference of such variates taking into account within subject correlation (possibly in R)?
If this is a simple subject correlation (i.e. like random intercepts in a mixed effects model), one can use a frailty model. An R-package to do this is "frailtypack".
If you wanted to fit more complex random effects, you can use the R-package "coxme" for fitting a Cox-PH model with general random effects.