In order to investigate the factors affecting the reproduction efficiency of Arab breeding mares in Tunisia, breeding data corresponding to 2340 mated cycles, over 5 years (from 2000 to 2004), from 555 mares offspring of 50 sires and bred with 17 stallions were analysed using a multivariate logistic regression. We chose logit link function and binomial distribution and we used log-likelihood-ratio (LL) and Wald tests (X(2)Wald) to test the mean values. The factors of variation included in the model were the year, the stallion, the age of the mare at mating, the sire and the category of the mare. Logistic regression results showed that age and sire affected both the overall conception rate and the first-cycle conception rate of the mares whilst category and stallion influenced only the overall conception rate. To our knowledge, this is the first study showing that the sire of the mare can affect both the first-cycle conception rate and the overall conception rate of the mares.