Classification of patient samples is an important aspect of cancer diagnosis and treatment.The support vector machine (SVM) has been successfully applied to micro-array cancer diagnosis problems.However, one weakness of the SVM is that given a tumor sample, it only predicts a cancer class but does not provide any estimate of the undcrlying probability.The penalized logistic regression has the advantage of additionally providing an estimate of the underlying probability that the sample will be assigned to each class, but it does not offer any probability of the class conditional upon an individual gene expression measurement.We propose the conditional logistic regression (CLR) model, which is an alternative method for the microarray cancer diag-nosis classification, and for the underlying probabilities of thc outcomes given an individual gene expression measurement.Furthermore, since the gene selection acts as a primary goal in microarray cancer diagnosis problem, we propose a new method called modified univariate ranking (MUR) as a new choice for dimension reduction.