53 Present address: 1 Scientist, Animal Genetics, Incharge LAR Section, IVRI, Izatnagar 243 122 (email: vetamitchandan @yahoo.co.in). 2 Head and Principal Scientist, Dairy Cattle Breeding Division. The genotype × environment interaction is the phenomenon that different environments do not equally affect performances of different genotypes (Falconer 1990). The analysis of variance techniques are not optimal for variance component estimation due to sire × herd (S × H) interaction that can arise from heterogeneity of residuals and (or) additive genetic variance among herds (Dickerson, 1962); from use of small number of selected sires; from differential non random mating among herds or from preferential treatments of some paternal half sibs. During breeding value estimation, if S × H interaction is not taken into account the interaction might bias EBV and reduce effectiveness of selection, because sires might rank differently across different environments. Selection for improvement in specific environment of the seedstock herd is thus warranted only if the environment of seedstock herd is more similar to the customer environments of commercial herds where germplasm is to perform, a situation that in most cases seems unlikely. The “Associated Herd Progeny Testing in Sahiwal” a multilocational programme was started in 1979 to bring about genetic improvement in Sahiwal cattle. The purpose of this study was to evaluate impact of sire × herd interaction for milk production in Sahiwal cattle using sire model of BLUP and 2 different animal models of DFREML. MATERIALS AND METHODS First lactation 305 days or less milk yield (FL305DMY) records of 903 Sahiwal cows, sired from 46 sires of same breed with 5 or more daughters, maintained at 3 different associated small herds namely National Dairy Research Institute (NDRI), Karnal, Government Livestock Farm (GLF), Chak Ganjaria, Lucknow and Government Cattle Breeding Farm (GCBF), Anjora, Durg over a period of 28 years (1979 to 2006) were analyzed after standardizing the records. Each of the sires had their daughters in at least two herds under the study. At 2 farms, namely GLF, Lucknow, and GCBF, Durg, the suckling was being practiced; whereas at NDRI farm weaning practice was followed which revealed about differences in managemental practices. The data structure is summarised in Table 1. Data were analyzed using sire model of BLUP (Henderson 1973) and 2 different animal models of DFREML described by Grasser et al. (1986) and Meyer (1989, 1998). The sire model and first animal model included fixed effects of herd, season, period and age at first calving as linear co-variable along with sires as random effect. In second animal model sire × herd interaction effect was included as additional random effect along with all the effects of first model. The sire evaluation method with lowest error variance was adjudged as most efficient and appropriate. Accuracy of the effect of interaction was tested by comparing log likelihoods of the model that included the sire × herd interaction to those that included only sire effects Indian Journal of Animal Sciences 79 (11): 1139–1142, November 2009 Effect of sire and herd interaction on milk production in Sahiwal cattle AMIT KUMAR 1 and R S GANDHI 2 National Dairy Research Institute, Karnal Haryana 132 001 India Received: 18 February 2009; Accepted: 29 April 2009 ABSTRACT First lactation records (903) of Sahiwal cows, sired by 46 sires of same breed maintained at 3 different farms were analyzed using sire model of best linear unbiased predictions (BLUP) and 2 different animal models based on derivative free restricted maximum likelihood (DFREML) algorithm. The animal model with sire × herd (S × H) interaction as additional random effect was found to be the best in terms of efficiency and accuracy over other two models. The animal model with interaction explained 15.2% of total phenotypic variation of milk yield due to sire × herd interaction and the estimate of heritability for milk yield was highest (0.182). The herd specific breeding values of sires estimated from interaction model ranged from –441.60 kg to 401.39 kg. The genetic slippage was noticed in most of the sires due to differences in managemental practices followed at different farms. The Karnal herd had highest breeding worth whereas lowest breeding worth was for Lucknow herd. The pooled breeding value over herds showed a declining trend over periods. Key words: Animal model, BLUP, Breeding value, Genetic slippage, Sahiwal, Sire × herd interaction