The aim of this paper is to derive the methodology for planning an optimal accelerated life test with the consideration of type-I censoring. In a typical industrial setting, the total duration of ALT tests must be controlled as failure times are random in nature. The generalized linear model approach allows optimal designs to be found using iteratively weighted least squares solution without directly calculating the expected Fisher information matrix, which is often intractable in the case of censoring. This approach is demonstrated with an assumed Weibull distribution. We discuss both D-optimal design, where the determinant of variance-covariance matrix of model parameters is minimized, and UC-optimal design, where the prediction variance of lifetime at a product's use condition is minimized.