Context The planning and execution of selection programs requires estimates of the genetic correlations between traits. As genetic change is achieved for a given trait, it is important to consider possible genetic changes for other traits. Understanding the magnitude and direction of genetic correlations can assist in selection decisions. Aims The aim of the present study was to estimate the genetic correlations of reproductive traits with productive traits and with percentages of fat and protein in the milk of dairy buffalo. Additionally, genetic trends were estimated for the traits under study over the years. Methods Data from 11530 complete lactations of 3431 female buffalo were used. The following traits were analysed: milk, fat and protein yields; percentages of fat and protein; age at first calving; and calving interval. The (co)variance components were estimated by Bayesian inference in multi-trait analyses, considering a linear animal model. To calculate the genetic trends, the average annual genetic values were regressed on the year of birth. Key results The means of genetic correlations estimated between reproductive (age at first calving and calving interval) and productive (milk, fat and protein yields) traits were positive, but of moderate to low magnitude. The association between the reproductive and milk quality (fat and protein percentages) traits were negative and of low magnitude. Genetic trends for the productive traits were positive (5.25 ± 0.63, 0.15 ± 0.034 and 0.09 ± 0.038 kg/year for milk, fat and protein yields respectively). Genetic trends for the reproductive traits of age at first calving and calving interval increased by 0.47 ± 0.09 and 0.48 ± 0.10 days/year respectively. In terms of milk quality, however, the percentages of fat and protein decreased by 0.016 ± 0.003 and 0.011 ± 0.001%/year respectively. Conclusions Genetic gains in productive traits may elevate the number of days at first calving and extend the calving interval, in addition to leading to the production of milk of lower quality. Implications The use of a multi-trait selection index is an alternative, as it combines information from different sources, such that an optimal selection criterion can be achieved over time by virtue of its emphasis on appropriate weighting for all traits.