Bivariate survival data arises when we have either a pair of observation times for each individual or times on two related individuals, such as infection times for the two kidneys of a person or death times of twins. Such data are also often subject to censoring - bivariate censoring - i.e., exact observations may not be available on one or both of components because of drop-out or other reasons. Hence it is important to have an efficient, nonparametric bivariate survivor function estimator under censoring, i.e., a bivariate Kaplan-Meier estimator. In this thesis we carry out an extensive simulation study of an estimator proposed by Sen and Stute (2007), which involves solving for an eigenvector of a certain matrix. A comparison of the estimator with two other existing but unsatisfactory ones is also given using a small data-set. Moreover, variance of the former is computed using a bivariate analogue of Greenwood's formula, which involves solving a matrix equation of the form AXB=C