SUMMARY1. Litters of mice, Mus mwsculus, of the highly inbred strain A2G/Tb, were reduced to four at birth and cross-fostered within and between three classes: (i) bred at 21°C (controls); (ii) the first generation reared at -3°C ('new stock'); (iii) the seventeenth to twenty-third generations reared at -30 C ('old stock'). There were therefore nine classes of fostered mice.2. There was a higher death rate in the nest and after weaning among mice of new-stock parentage, regardless of foster-parentage. 3. Litters reared at 210 C were heavier at 3 weeks than those reared at -30 C, regardless of parentage; the effect of temperature was also evident in body weights at 16 weeks. There was compensatory growth between 3 and 16 weeks, shown by lower variance in body weight, within classes, at 16 weeks.4. Members of fostered litters were mated, and their reproductive performance recorded to the age of 28 weeks.5. More young were born and weaned per pair at 210 C than at -3°C, regardless of true parentage. In contrast, nestling mortality depended principally on true parentage; there were fewer deaths among the young of mice whose true parents were old stock, regardless of foster-parentage; this effect was especially evident for losses of whole litters.6. Mean body weights of the nine classes of fostered females were positively correlated with mean numbers of young born to them, and with the mean weights of their young at 3 weeks; but there was no such correlation with death rates among their young.7. Young of some of the fostered mice were also mated. The effect of old-stock ancestry on nestling mortality was not evident in this further generation. Some mice were studied after transfer between temperatures without fostering. Their breeding performance confirmed that old-stock mice did not retain their superiority after two generations at 210 C.