Data were analysed from a long-term selection experiment with Merino sheep, based on immunological responsiveness to the intestinal nematode Trichostrongylus colubriformis. For the first 14 years of selection, the criterion was the mean of five fortnightly faecal worm-egg counts (FECs) of pen-housed lambs that were vaccinated with irradiated larvae then challenged with normal T. colubriformis larvae. For most of the lambs born in the subsequent 6 years, the selection criterion was the mean of three weekly FECs of grazing lambs following a secondary challenge with T. colubriformis larvae. Data from 2233 lambs were included in the analyses.

At the end of the experiment, the lines selected for high and low response to challenge differed by 2·3 to 2·9 phenotypic standard deviations. The heritability of average pen-tested FEC was 0·38 (s.e. 0·04), similar to that for average FEC after secondary field challenge (0·37, s.e. 007). Average FECs from the primary field challenge were less heritable (0·21, s.e. 006). Counts recorded on pen-tested animals at 3, 5, 7, 9 and 11 weeks after infection were all highly correlated genetically (estimates 0·93 to 0·99) and of similar heritability (0·33 to 0·39) with a non-significant tendency for later FECs to be more heritable. Phenotypic correlations were lower, ranging from 0·60 for counts 8 weeks apart, to 0·78 to 0·81 for adjacent samples 2 weeks apart. Single-record heritability estimates at 3, 4 and 5 weeks after secondary infection in grazing lambs did not differ statistically from each other but were highest at the 5th week after infection at 0·33 (s.e. 007).

FECs determined in pens were imperfectly correlated with FECs determined at pasture (genetic correlation 0·72, s.e. 013). Phenotypic correlations of single-record FECs across primary and secondary field challenges were low (around 0·2), although the genetic correlation between mean primary FEC and mean secondary FEC was not significantly different from unity. Testing under standardized conditions in pens did not result in consistently higher heritabilities than testing immunologically primed lambs at pasture. Repeated measurement in a breeding programme offers little additional benefit, except when lambs have had ample prior experience of parasite infection, with the two measures separated by an anthelmintic treatment. If an objective of a Merino breeding programme is to reduce FEC under field conditions, then using one measure in the field, approximately 3 to 5 weeks after a secondary artificial infection, is likely to lead to long-term progress.