Methods for detecting influential observations for the Weibull model fit to censored data are discussed. These methods include: one-step deletion diagnostics, influence functions and curvature diagnostics. Results indicate that the curvature diagnostics may be helpful in detecting masking.