Mice of three kinds have been bred continuously for a number of generations in a cold environment. The mice were (a) of two inbred strains (A2G and C57BL) and (b) of a hybrid strain derived from four inbred sources. Parallel lines of each strain were bred in two rooms kept respectively at -3 (experimental) and 21 degrees C (controls). At -3 degrees C the hybrid mice were systematically selected for fertility. No other deliberate selection was practised. The hybrid mice were at all times more fertile than the inbred mice. Mice of any given kind were less fertile in the cold environment than those of the same kind in the warm. The number of young born per pair to the hybrid mice increased over twelve generations in both temperatures; there was no such change in the inbred strains. At the twelfth generation the hybrid mice in the cold were heavier at 3 weeks and at 16 weeks of age than they had been at the sixth; at 16 weeks they were heavier than the controls. Some twelfth-generation hybrid mice were transferred to the warm environment, and their offspring were heavier still. The weight of the hybrids' hair coat was higher in the cold than in the warm, but transfer of mice from cold to warm showed that this was a direct effect of the cold and not due to genetical change. A2G and hybrid mice, reared to the age of 5 weeks in the warm and then transferred to the cold, were more fertile and provided a better nest environment (that is, probably milk supply) than those of the same strains reared in the cold. This probably held also for C57BL mice, but many of these were barren after transfer to the cold. At first, mortality before weaning at 3 weeks was higher in the cold than in the warm, but in all three kinds of mice there was a progressive reduction in nestling mortality in the cold. This decline was spread over nine generations in the A2G mice, eight generations in C57BL and twelve generations in the hybrids. No such improvement was shown by the control A2G or hybrid mice; but the C57BL controls showed some decline in nestling mortality over sixteen generations. This cannot be attributed to a genetical change in the inbred mice, and was probably not a genetical effect in the hybrids. It is suggested that it could be due to either (i) a cumulative maternal effect operating on a previously unsuspected scale; or (ii) to a progressive decline in the action of a pathogenic organism.