The aim of this research was to model the lactation curve in dairy cows with nonlinear models (MNL) and nonlinear mixed models (NLMM), in an intensive breeding dairy farm during the period 2007-2013. In total, 22 173 data referred to biweekly milk production controls were analysed, corresponding to 1302 lactations and 668 Holstein cows. The Brody, Wood, Cobby and Wilmink mathematic models were used evaluating NLM and NLMM, considering the cow as a random effect using SAS v. 9.4 statistical package. Applying information from Akaike (AIC) and Bayesian (BIC) method, NLMM was found to be the best fit for the 305-day lactation curve. Wood mixed model was the best fit for the first calving and Wilmink mixed model for the other calvings. Likewise, production peak, days to production peak, percent productive yield by thirds during the lactation period and production at 305 days according to the number of calving were determined.