An animal model was used to estimate heritabilities of faecal egg count (FEC) and packed cell volume decline (PCVD) in lines of Merinos selected for divergent levels of resistance to Haemonchus contortus. The estimated heritability of FEC was 0·23 (s.e. 0·03) and of PCVD was 0·21 (s.e. 0·03). A cube root transformation was found to be effective in normalizing FEC data and reducing the range of within selection line-birth year variances from 118-fold to 10-fold. Transforming FEC data increased the heritability estimate to 0·29 (s.e. 0·03) and reduced selection bias due to heterogeneity of variance, but the further step of pre-adjusting the transformed data to a common variance had only a trivial effect. Heritability estimates for FEC in years with high means and variances were similar, as were estimates for transformed FEC. The estimated genetic correlations between measurements in years with high and low means and variabilities were 0·77 to 0·91.
PCVD and pre-adjusted transformed FEC had a phenotypic correlation of 0·48. Estimates for the genetic correlation ranged from 0·76 in the decreased resistance line to 1·00 in the increased resistance line and 0·87 with pooled data. Other effects on resistance included birth type and dam age, with twins and progeny of maiden dams being more resistant than their cohorts. Sex effects were unimportant for FEC but males had higher PCVD than females in most, but not all, years. Younger animals had higher FEC and PCVD than older animals. Maternal genetic effects were found to be unimportant, as were the effects of low levels of inbreeding.