PROC LIFETEST is a nonparametric approach to the analysis of survival
data, meaning it makes no assumption about the underlying
distribution. To quote SAS online documentation: "You can use the
LIFETEST procedure to compute nonparametric estimates of the survivor
functions, to compare survival curves, and to compute rank tests for
association of the failure time variable with covariates."
You mention you have groups. If your 'groups' are treated as strata,
you can test for a difference in SURVIVOR FUNCTIONS between the strata
if you use the following code:
proc lifetest data=yodata plots = (s) graphics;
time timeVAR * censVAR(0);
you will get to see a neat survival function plot, a log-rank test of
equality of survival function over strata.
However, to test for which distribution fits your data best (log-
normal, exponential, log-logistic, weibull) one option is the LIFEREG
procedure. For an example of testing model fits, please see [VA Lung
Cancer Data: Example 2] at http://tinyurl.com/2dfpdm
While you are at this, be sure look into model nesting (e.g.
exponential is a special case of Weibull) and try to fit the most
parsimonious of several competing models.