Log-rank tests are sometimes used to analyse longevity data when other tests should be preferred. When the experimental design involves more than one factor, some authors perform several log-rank tests with the same data, which increases the risk to wrongly conclude that a difference among groups does exist and does not allow to test interactions. When analysing the effect of a single factor with more than two groups, some authors also perform several tests (e.g. comparing a control group to each of the experimental groups), because post hoc analysis is not available with log-rank tests. These errors prevent to fully appreciate the longevity results of these articles and it would be easy to overcome this problem by using statistical methods devoted to one-way or multi-way designs, such as Cox's models, analysis of variance, and generalised linear models.