The initial dataset included a total of 7819 monthly test-day records of 1205 buffaloes located in six different farms and recorded by Cattle Information System/Egypt (CISE). The buffalo having less than five test-day records were removed from the dataset. The test-day records after 300 days of lactation were excluded from the analysis. Also, the extreme phenotypic values of daily milk yield, fat and protein percentage were removed from the dataset. Three different models were tested for fitting the lactation curve of the Egyptian buffalo, Wilmink (WM), Guo and Swalve (GSM) and Wood (WD). The expected daily milk yield at the beginning of lactation was 8.90, 9.16 and 9.08 kg in WM, GSM and WD, respectively, and then increased to reach the peak 10.56, 10.67 and 9.08 kg in WM, GSM and WD, respectively. Days to attain the peak were 29, 26 and 28 day, in the three models, respectively. The daily milk production was decreased to 2.41 kg in WM and WD and 2.64 kg in GSM at the end of the lactation period. The curves of milk fat, protein and lactose yield had similar pattern to the curves of the daily milk yield. The milk fat and protein percentage seem to have an increasing trend with the increasing of days in milk. The curve of milk lactose percentage was in contrast with the curve of the milk fat and protein percentage. The three criteria used to compare the goodness of fit of these models were Mean Square Prediction Error (MSPE) value, Akaike Information Criterion (AIC) and Schwartz’s Bayesian Information Criteria (BIC). The goodness-of-fit statistics of the expected curves for daily milk yield and its contents, for the better-fitting models, appear that WD model gave the best fit for the studied criteria.
