This study was designed to screen a large number of sheep to identify individuals with high and low methane (CH4) production, and to estimate repeatability and heritability of CH4 emissions in sheep, utilising portable accumulation chambers (PAC) designed for in-field use. Mature ewes (n = 710) selected from a research flock with known sires had their CH4 production over 1 h measured in PAC [CH4 (g1h)]. Individuals with High (n = 103) or Low (n = 104) CH4 (g1h), adjusted for liveweight (LW), were selected and re-measured on three occasions 1–4 months later, at another site with more abundant and better quality pasture. Mean of the selected (207) ewes CH4 (g1h) emissions were ~50% higher than at the first measurement site (0.66 g vs 0.42 g). LW was a significant correlate of CH4 production (r = 0.47). Correlations between CH4 (g1h) for the three PAC measurements at Site 2, before adjusting for LW ranged from 0.44 to 0.55. After adjusting for the effect of LW, repeatability was 0.33 at the first and 0.43 at the second site. The correlation between estimates of an animal’s emissions at the first and second sites, adjusted for LW, was 0.24. Initial CH4 production of the selected High group was 32% greater than the Low group (P < 0.0001). On re-measurement there was still a significant difference (9–15%, P < 0.006) between Low and High groups. The initial estimate of heritability of CH4 (g1h), based on variation between the ewes’ sires (0.13), was not maintained across the two sites. This may be due to genotype × environment interactions. We postulate that aspects of rumen physiology, which modulate CH4 production, could be expressed differently in different nutritional environments. Our results indicate that field use of PAC to screen sheep populations for CH4 production is both robust and repeatable. However, further investigations are required into the relationship between CH4 output of individual animals in PAC compared with the more controlled conditions in respiration chambers.