The aim of this study was to assess the suitability of a static neural network (SANN), a MLR model, and a nonlinear auto regressive model with exogenous input (NARX) for the prediction of total daily herd milk yield (DHMY) over varying forecast horizons. The most successful model was selected according to its abilities to generate the most accurate forecast using very limited training data in low volumes over a long- (305 d), medium- (30 to 50 d), and short-term (10 d) horizon.