In this paper we derive a law of the logarithm establishing the exact rate of uniform convergence for nearest neighbor kernel estimators of the density and hazard rate functions in the presence of censoring.