Variance components of clinical mastitis in dairy cattle — effects of trait definition and culling

First lactation clinical mastitis records for Norwegian cattle from 1978 onwards were analysed. Variance components for clinical mastitis were estimated with a linear sire model using records of more than 500,000 daughters of 2043 sires. Heritability increased slightly as the period for sampling health data increased, and the longest period analysed (from 15 2 days before calving to 210 days after calving) gave the highest heritability estimates (h 5 0.04). However, a sampling 2 period from 15 days before calving to 30 days after calving captured most of the genetic variation (h 5 0.03), and showed a high genetic correlation (.0.94) with clinical mastitis sampled over a longer period of first lactation. This implies that recording of clinical mastitis over a short time period around first calving can provide a measure of clinical mastitis with a substantial value in genetic evaluation. Using culling reason as an additional source of information about mastitis increased heritability only slightly compared with using clinical mastitis records only. Excluding cows culled before the end of the sampling period from the data if they have not had mastitis resulted in a higher heritability of mastitis than both including a fixed effect to account for culling in the model and a bivariate analysis of clinical mastitis and culling. Hence, culling affects variance component estimates of clinical mastitis.  2001 Elsevier Science B.V. All rights reserved.