Mice were selected for one of three criteria: appetite (A), measured as 4- to 6-week food intake, adjusted by phenotypic regression to minimize change in 4-week body weight, fat percentage (F), using the ratio of gonadal fat pad weight to body weight in 10-week-old males, and total lean mass (protein, P), using the index, body weight in 10-week-old males − (8 × gonadal fat pad weight). For each selection criterion, there were 3 high, 3 low and 3 unselected control lines. At generation 11, the high and low A lines diverged by 17% of the control mean and the realized heritability from within family selection of adjusted food intake was 15%. Selection for this character produced changes in body weight, gross efficiency from 4 to 6 weeks, and percentage of fat, the high lines being heavier, more efficient and less fat than the lows. The high and low F lines diverged by 80% of the control mean and the realized heritability of the ratio of gonadal fat pad weight to body weight was 44%. Selection for this character produced changes in total fat per cent, but little change in percentage protein, body weight, food intake or gross efficiency. The high and low P lines diverged by 40% of the control mean and realized heritability of the lean mass index (10-week weight − [8 × gonadal fat pad weight]) was 51%. Selection for an increase in the index increased body weight at all ages, food intake and 4- to 6-week gross efficiency. There was no change in percentage fat. Responses in the selected traits were not highly correlated, and the different lines provide an opportunity for investigating responses in physiology, metabolism and gene products.