In this paper are proposed several statistical measures for diagnostic and validation of the new class of generalized Weibull linear regression models introduced by Prudente and Cordeiro (2010). This class of models is one more alternative for the analysis of positive continuous and skewed data, which due to its properties is competitive mainly to models with gamma response. Are proposed two residuals and its statistical properties are studied analytically and through Monte Carlo experiments. Some measures are also derived based on case-deletion and mean-shift outlier models such as one-step approximation, likelihood displacement, generalized Cook’s distance and outlier’s tests. A test for assessing the assumption of homogeneity of coefficient of variation is also derived. Finally, an application on advanced inoperable lung cancer patients is give to illustrate the residuals and diagnostics tools derived. Models with gamma and Weibull responses were fitted to the data. The analysis suggested that Weibull model fit the data better.