The logistic regression model is remarkably flexible. Nevertheless, lack of fit may also occur so to be able to detect departures from the model is important. In this paper, a generalized empirical probability plot based on generalized residuals is developed. The method is illustrated with simulated and real data.