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.Keywords: Czech White Shorthaired Goat; fat; protein; lactation curve; test day 466Original Paper Czech J. Anim. Sci., 49, 2004 (11): 465-473 day Animal Model for BVE in dairy goats. Milerski et al. (1999) found considerable differences in the evaluation of individual animals between the system of evaluation of dairy goats presently used in the Czech Republic and BVE done by BLUP Animal Model. MATERIAL AND METHODSData from the official milk recording of the Czech White Shorthaired breed of goats provided by the Sheep and Goat Breeders Association (SCHOK) was used as the basic material for our calculations. The analysed data were recorded in the period from 1992 to 2002. Daily milk production was recorded by methods AT, AC or A4 (ICAR, 2001). Although the analysed data were recorded in production operations, several fixed effects were included in the model. The most marked effects will be chosen using the ordinary Least Squares Method and the procedure GLM (SAS, 1996).Several checks of performance records were carried out in order to exclude logical inconsistencies and biological improbabilities: (a) Age at first kidding had to be at least 275 days. (b) Difference between lactation number and period (in years) from birth to the 31/12/n (n = production year): 0.5-2 years. (c) Test-day records included for year n had to be between the dates 1st October n-1 and the 31st March n+1. (d) Kidding date had to be between the dates 1st October n-1 and the 30th September n. Other used criteria are shown in Table 1. Poor quality data were excluded from the followings computations. The original performance record database included 87 736 test-day records. The number of data considerably diminished a�er the introduction of the above-mentioned criteria, a total of 78 736 records of daily milk production, 77 536 records of fat percentage and 78 269 records of protein percentage in milk from 6 234 goats were finally analysed.The effects of herd, test day, number and stage of lactation, production year, kidding season and li�er size were tested using the followings five models: e ijklm = random residual RESULTS AND DISCUSSIONMeans and standa...