Genotype by environment interactions (G × E) can form a potential source of inefficiency in animal breeding if selection decisions are made without acknowledging their effects. The presence of such interactions between two Scottish hill farms was investigated using performance data collected from 18,459 Scottish Blackface lambs between 1997 and 2010. Pedigree information was available for 27,548 animals. 30 out of 379 sires used during this time period were used on both farms. Farm A is located on the East Coast of Scotland, whereas Farm B is on the West Coast. The farms differ in a number of aspects including annual rainfall, topography, vegetation, temperature and altitude ranges, with Farm B representing a harsher environment overall. Traits studied were birth weight, 8-week weight and weaning weight, ultrasound back-fat and muscle depths at weaning, carcass weight, carcass fat grade and carcass conformation score. Genetic correlations were estimated for each trait, between the two farms, with those significantly different from 1 (P < 0.05) indicating the presence of G × E. The models used fitted relevant fixed effects as well as direct and maternal genetic and permanent environmental random effects. Correlations estimated were not significantly different from 1 for all traits apart from birth weight, which had a correlation of 0.45 (s.e. 0.31). By taking G × E into account and comparing bivariate and univariate analyses, the maximum selection response observed for birth weight was 0.0016 kg per generation. Overall, the lack of G × E observed for the majority of traits studied suggest that the common sire offspring have performed similarly across both farms. However, the presence of G × E associated with birth weight may have implications for lambing associated problems, or lamb survival, if sires produce lambs with unexpectedly high or low birth weights. © 2011 Elsevier B.V. All rights reserved