Abstract: The objective of this study was to reveal the lactation curve and the influencing factors for large dairy cattle farm in south of China. The lactation curve was fitted by 5 mathematical models (including the wood’s incomplete gamma function model, Nelder inverse quadratic polynomial model, Wilmink model, mixed log model and AliSchaffer model) and 4 time interval (including daily, weekly, ten days and milking monthly) on 296 895 daily milk yield records of 980 Chinese Holstein with all first lactation to fifth lactation from year 2009 to 2010 in Haifeng Dairy Farm. The effects of models, time interval and parity on fitness and parameter of lactation curve were analyzed by ANOVA. The results showed that the fitness of 5 lactation curve models ranged from 0.839 7 to 0.954 1. The models and parity showed significant effect on the fitness of lactation curve at P<0.01, and the time interval showed no significant effect on the fitness of lactation curve. By the fitness of lactation curve, the AliSchaeffer was largest, and the Nelder inverse quadratic polynomial model was least. The fitness of lactation curve for the first lactation was higher than that of the third lactation for all models. The a value of lactation curve was affected significantly by models, time interval and parity at P<0.01 level. The b value of lactation curve was affected significantly by models at P<0.01 level, and by time interval at P<0.05 level. The peak milk yield and the day in peak milk yield were affected significantly by the models and parity at P<0.01 level. The result indicate that the AliSchaeffer model is the best model to fit the lactation curve for Haifeng Dairy Farm based on all information in this study.