Correlated responses in caudal vertebrae number (VN), lengths of eighth and ninth caudal vertebrae (V8 and V9, respectively), femur length (FL) and femur weight (FW) were evaluated in lines of mice which had been selected for six-week body weight (WK6) and/or six-week tail length (TAIL). Ten males and ten females were randomly sampled from each of ten selected lines (two replicates each of five selection treatments) after seven generations of selection. Sexes and lines were significant (P < .01) sources of variation in all seven traits. Sex x line interactions were unimportant except for V8 and V9. Male mice of both replicate lines selected for increased WK6 and decreased TAIL had shorter vertebrae than females, whereas the reverse was true for all other lines. Multiple regression and canonical correlation analyses indicated a high phenotypic relationship of FL with both WK6 and TAIL. Examination of the correlated responses indicated that FL was the only skeletal trait that showed a substantial correlated response to single trait selection for both WK6 and TA IL. Thus, the genetic relationships among the three traits also appeared to be high. Between replicate variation was not significant for randomly selected control lines. However, about onethird of the statistical tests between selected replicates were significant. This was taken to indicate a joint effect of selection and drift in causing variation between replicate lines. Replicate variation was further examined by canonical variate and generalized distance analyses. The first two canonical variates accounted for most of the generalized variance. Graphically, the first two canonical variables discriminated among selection treatments, whereas the replicates tended to cluster. Thus, although between replicate differences were significant for several traits, the differences were relatively small compared with the variation between lines having different selection criteria.