In this thesis, we consider a regression models for survival censored data of type II in which the underling distributions are exponential or gamma where the effect of the regressor variables on the means is multiplication given by the model Three methods of estimation for the regression coefficients are considered, namely maximum likelihood (ML), weighted least squares (WLS), and suggest weighted least squares (SWLS). These methods are discussed theoretically and examined practically by Monte Carlo simulation for the case of a single explanatory variable. Moments and higher moments properties of the estimators, such as, bias, variance, skewness, and kurtosis are examined, illustrated and compared. Finally, a new bias reduction estimator to the ML estimator is proposed and shows a higher performance with respect to the other estimators.