The objective of the present study was to apply principal component analysis (PCA) on Brazilian Dairy Herd Improvement (DHI) data to discover the subset of most meaningful variables to describe complete lactations. The Holstein Livestock Breeders Association of Minas Gerais provided data collected between 2005 and 2016 from 122 dairy farms located in the State of Minas Gerais – Brazil. Twelve numerical variables were selected from the original dataset and four additional variables were created. The final dataset contained 28379 observations of 16 numerical variables. They were entered into a Pearson correlation matrix and highly correlated variables (r > 0.94) were evaluated for exclusion based on biological relevance. The PCA was performed on selected variables (n = 12) after they were standardized to mean = 0 and standard deviation = 1. Five variables were PCA-selected as meaningful to describe the variation of complete lactations. They were age at calving, lactation number, milk yield on first test day, energy-corrected milk, and total solid yield on 305 days of lactation. These variables could be used to evaluate complete lactations and future work using Brazilian DHI metrics could focus on modeling the relative importance of each of the selected variables.