Graphical methods for testing goodness of fit are reviewed and classified as transformations of P-P and Q-Q plots. Survival and cumulative hazard plots for censored data, probability plots, and generalized residual plots are included. Difficulties in interpretations arise when the points are grouped more closely together in particular areas of the plot. This problem commonly occurs in cumulative hazard and Q-Q plots, especially in large data sets, but can occur in P-P plots when censoring is present. An empirical rescaling of the axes is proposed to overcome this problem. The techniques are applied to reliability and clinical trial data sets.