The purpose of this research has been to model the milk yield, fat content, protein content and liveweight of Irish dairy cows over a lactation period. The analysis was carried out on 15,729 lactation records from commercial and experimental herds including both autumn and spring calving animals of various breeds. Initially, the factors which affect milk yield, its constituents and liveweight were investigated and then a scientific method for detecting abnormal recordings was formulated. This method of detection is very effective and abides by the guidelines outlined by the International Committee of Animal Recording. After removing the abnormal recordings from the data, a number of models were fitted and their suitability was assessed on the basis of their good ness-of-fit and adherence to the assumptions made in carrying out regression analysis When modelling milk yield, a severe problem of multicollineanty was encountered and methods of reducing multicollmearity were investigated. As a result a new model was developed, which provided an acceptable level of accuracy in representing the shape of the lactation curve for Irish dairy cows. A model that satisfied the assumptions of regression analysis and predicted the actual content of the constituents to within 0 01 per cent of the actual values was also developed while a novel approach was used for modelling liveweight Firstly, splines were used to find the dimensions of the data and principal component regression was used to estimate the regression coefficients of this new model. This model satisfactorily represented the shape of the hveweight curve and it can be easily updated for use by bio-economists. The models proposed in this thesis are currently being incorporated into the Moorepark Dairy Systems Model which investigates the challenges that currently face the Irish dairy industry.