ABSTRACT: The aim of the paper was to prove the accuracy of various statistical models of variance analysis for estimation of systematic factors that influence milk yield and fat and protein content in dairy goats. Data on daily milk production recorded by methods AT, AC or A4 for the population of Czech White Shorthaired Goats over 1992?2002 was used for calculations. A total of 78 736 test day from 6 234 goats were analysed. Average daily milk production was 3.09 kg with 3.72% of fat and 2.84% of protein. Variability of examined traits was significantly influenced by effects of herd-year or herd-test day, li􀄴er size, parity, season of kidding, stage of lactation and the combined effect parity-year-season of kidding. The variants of model equations in which complex effects of herd-year were replaced by effects of herd-test day showed higher values of reliability for the three traits analysed.