Mass selection for feed conversion efficiency (FCE) in mice was practiced for six generations in two lines fed ad libitum (AL) and two lines restricted to 80% of ad libitum (R). In one line of each pair of selected lines and test period was terminated after a fixed time (14 days) (TAL and TR) and in the second line of the pair after consumption of a fixed time (14 days) (TAL and TR) and in the second line of the pair after consumption of a fixed quantity of feed (WAL and WR). A random mating control line (C), also fed ad libitum was maintained throughout the study. Correlated traits measured were initial weight, final weight, weight gain, feed consumption and litter size. A significant response in FCE was found in the TR and WR lines, the regression of response on generation number being -1.36 +/- 0.31 and -1.09 +/- 0.39 respectively. There was a significant decrease in initial weight on test in the TR and WAL lines and in those lines mature body weight was significantly lower than in the C line when measured in the sixth generation. Heritability of FCE was higher in the restricted line (0.31 +/- 0.12 for TR and 0.37 +/- 0.29 for WR) than in the ad libitum lines (0.13 +/- 0.20 for TAL and 0.13 +/- 0.25 for WAL). Heritability of FCE was 0.17 +/- 0.31 in the C line. Phenotypic and genetic correlations between FCE and weight gain were high and negative in all lines. Progeny from each line were tested in each of the other feeding regimes each generation to measure lin X feeding regime interaction. This interaction was statistically significant.