One of the important statistical problem in scientific studies is determining the factors that are mostly affecting the concerning phenomenon or in which level these factors have the highest effect. CHAID (chi-squared automatic interaction detection) analysis was developed as a method which determines the combinations or interactions among the variables. The method, is a technique that repeatedly partitioning the population into different subgroups or segments where the variation of dependent variable is minimum within the groups and maximum between the groups. In this study, compound categories and subgroups of independent variables for obtaining the maximum milk yield were estimated by CHAID analysis where 2xMEx305 milk yield is used as a dependent variable and lactation duration, dry period, age at first conception, calving interval, the number of insemination per conception, service period, calving season, gestation period are used as independent variables in 440 Holstein cows grown at the Bala State Farm in the period of 1985-1992. Dry period for 55-74 days was estimated as the first subcategory for the best explaining of the milk yield among the independent variables which are accepted as affecting the milk yield. Age at first conception for 10-16 month was determined as a subcategory that is possessing the maximum interaction with dry period of 55-74 days. The Answer Tree 1.0 package program was used for applying this method.