There is a growing number of publications on Technical Efficiency (TE) of dairy farms, due to the political, economic and social relevance of the theme as well as to its academic appeal. The goal of this thesis is two fold: first, it presents a research synthesis on TE in dairy farms conducting both a meta-synthesis and a meta-regression analysis and, second, an empirical study aiming at measuring and understanding the factors affecting the TE in dairy farms of selected municipalities in the South of Brazil. The research synthesis embraces 103 papers published between 1987 and 2014 in 54 sources using a rigorous and reproducible approach to select studies. This synthesis complements and extends previous literature reviews on TE in dairy farms, offering a synthesis framework (i.e., context and determinants, inputs, outputs and measurement techniques of TE) and analysing the effects of different methodologies and study-specific characteristics on Mean TE (MTE) through meta-regression. Findings point to no agreement among authors on context and determinants of TE. The analysis of determinants led to verifiable research propositions: the main determinants of TE are geographical location (altitude, quality of the soil, climate), farm size, investments in veterinary care, feeding and milking practice, TE model estimation techniques, public policy, and management-related variables. Results from the meta-regression analysis of the gathered papers totalling 531 TE observations in dairy farms worldwide are discussed and makes two important contributions: (i) it updates and compares previous works on frontier estimation of TE in dairy farms and (ii) it adds two dimensions of dairy farms, size (herd and land area) and economic development, to the known differentials of TE measurement. The variation in the MTE indexes reported in the literature can be explained by the methodology of estimations (method of estimation, functional form of frontier models, model dimensionality), farms geographical location and farm size. Additionally, the results suggest that, given the state of technology prevailing in each region at the time that the studies on TE were conducted, dairy farmers in the sample could increase milk output by 20.1 percent (level of inefficiency), on average, if they produce on their frontiers. Implications for dairy farmers and for future research close the research synthesis. The propositions and hypotheses emerged from the research synthesis and were tested using data obtained in a survey conducted in selected municipalities in the South of Brazil. The mensuration of level TE combined with benchmark analysis helps the producers improve the degree of efficiency. This empirical study measured and helped in the understanding of the main factors affecting the TE in dairy farms from these municipalities, shedding new empirical lights in the academic literature.