Abstract: Bivariate (BV) analysis of Body condition score (BCS) jointly with milk yield was investigated as a method for BCS genetic evaluation and compared to univariate (UV) analysis of BCS records, adjusted or unadjusted for milk yield (n = 38 930). Additionally, milk yield was analysed with and without adjusting milk yield for BCS. An animal model with additive genetic and permanent environmental effects was fitted to data and analysed by REML. For BCS, the range of heritability (h2) was 0.21 - 0.26, which was similar to milk yield, showing the potential for genetic improvement. The magnitude of h2 and ?pe2 was smaller for adjusted than unadjusted BCS data for milk yield. Similar results were found for milk yield, with or without adjustment for BCS. This reduction in the genetic variation is a result of removing the differences in milk yield level / BCS. The UV and BV methods differed greatly when sire in data had small progeny group size (n=10). The BV analysis produced higher heritabilities and wider range and good accuracy of estimated breeding values for BCS showing that genetic response to selection is likely to be higher when it is based on multi-trait evaluations. Based on this result, a bivariate or multi-trait genetic evaluation with milk yield is recommended for BCS.