It is customary to use chi-square tests to compare survivals. However, if it was known that the treatments are ordered (for example different levels of a single treatment), the chi-square test is no longer appropriate. In this paper, we introduce an alternative technique to test the hypothesis of ordered survivals. We suggest a test statistic, derive its mean and variance and show that its asymptotic distribution is normal. The theorical results are illustrated through an application on real data sets (sources are cited in the text) for bladder cancers patients and another data set on treated and untreated cancer patients.