The productivity indices of dairy farming in Brazil are low when compared to countries with technologically advanced dairy farming. Productivity can be significantly improved through the use of animals with higher genetic value for milk production and appropriate management techniques. Worldwide, the Holstein breed is predominant for milk production, being also the most used in Brazil, both as a pure breed and in crosses, as a contribution to the development of the Girolando breed. In this project, records of productive performance and linear classification were used to carry out genetic evaluations and make the results available in Summaries of bulls, which are publications designed to guide the selection processes by breeders. The records from the databases of both breeds were also used, individually and jointly, in studies developed with institutional partners, aimed at improving the procedures applied in the respective genetic evaluation systems. Accordingly, studies were carried out on applying new methodological procedures to adjust the productions of the test-day milk yield (PDC) through random regression models, with polynomial and autoregressive functions. Other studies included the statistical modeling of lactation milk production, individually by breed and in multi-trait analyzes for the evaluation of Girolando bulls. During the project period, genetic evaluations were carried out for the production and conformation traits of the Holstein breed and the milk production of the Girolando breed. These results were published annually in the Embrapa Documents series and made available to stakeholders in the dairy production chain. Regarding the studies concerning the Holstein breed, the genetic parameters for test-day milk, fat and protein yields were estimated with auto-regressive models. The overall results suggest that it's feasible to make the transition from using the lactation milk production to the statistical modeling of PDCs which make it possible to obtain higher reliabilities in the prediction of breeding values of Holstein bulls and cows. In a similar way, other studies were carried out using the test-day records of the first lactation of Girolando cows to estimate covariance components by random regression, with Legendre Polynomials (LP) of order three to five. Heritability estimates were similar among models, ranging from 0.25 to 0.38, with the highest values observed at the extremes of lactation. In general, the obtained estimates suggest that the modeling of PDC by random regression with Legendre polynomials could replace the use of the lactation milk production in genetic evaluation of the Girolando breed. In another study, with autoregressive models (AR), using PDC of the first three lactations, heritability estimates of milk production ranged from 0.17 to 0.27, confirming the potential use of PDC to replace the lactation milk production for the prediction of genetic values of Girolando bulls and cows. However, it is important to evaluate the impact of these two models, comparatively, in the ranking of bulls and cows, reliability of predictions and potential for genetic progress in milk production of the breed. In yet another study, analyzes were performed using milk production records in the first lactation of the Girolando breed jointly with that from the Holstein breed. The objective was to identify and quantify the potential contribution of Holstein information to Girolando, through the use of common bulls. The heritability estimates were 0.22 for the Girolando analysis and 0.25 and 0.24 for the Girolando and Holstein breeds in the joint analysis. The genetic correlation between the milk yields of the two breeds was 0.73. The average accuracy of breeding values of 31 Girolando bulls predicted with the two analyzes were, respectively, 0.54 and 0.49 and the rank correlation between them was 0.10. These results did not indicate potential gains from jointly modeling the lactation milk production for the Girolando breed, although some changes were observed in the ranking of bulls. The interpretation of this particular result suggests a more specific analysis of the distribution of progenies of bulls to characterize whether its cause/origin is related to the structural nature of the available data sets. The results of all these studies were presented at national and international conferences, and thus published as abstracts in the proceedings of these events.