Six mathematical functions; Wood, Mitscherlich* exponential, Michaelis-Menten* exponential, Logistic* exponential, Weibull* exponential and Multiphasic; were used to describe the lactation curve of the monthly test- day average of 12020 Holstein Fresian milk records in the Kingdom of Saudi Arabia. The fifteen monthly test- day averages were 10 fitted using nonlinear regression by the Gauss-Newton iterative method. Measures of Goodness of fit such as Autocorrelation, Durbin Watson statistic (DW), R², and Mean square error were used to compare between different functions. Autocorrelation values ranged from .14 to .28 for all models. Durbin Watson statistic ranged from 1.062 to 2.785, R² values being greater than .98. Mean square error was the lowest for triphasic function (MSE=.1) and the highest for Mitscherlich* exponential and Logistic* exponential (MSE = 7.792). The Triphasic function with nine parameters was the most accurate function for representing the lactation curve. However, Mitscherlich* Exponential, Logistic * Exponential and Monophasic functions displayed the poorest fit for the data.