A new test is proposed for testing the validity of the logistic regression model based on case-control data. The proposed test does not need a partition of the space of explanatory variables to handle the case of nonreplication. The new test is consistent