The timing of life-history traits may have strong influences on the evolution of life cycles and on population demography. This is especially true of the age at which females first reproduce (Cole's principle). We examined whether the age at which females first reproduce influences fitness in Columbian ground squirrels (Urocitellus columbianus), for which females varied in the age at which they initially produce weaned offspring, from ages 1 through 5 years. With 148 females with complete known life spans in a 28-year data set, we examined four fitness measures: individual fitness (λ_ind), individual fitness relative to the pattern of growth of the population (λ_rel), lifetime reproductive success (LRS), and LRS relative to the total LRS for each female's cohort (LRS_rel). These metrics were calculated for offspring produced at the time of weaning and offspring that survived to emerge after their first hibernation period. Individual fitness (λ_ind) was significantly associated with population growth during a female's lifetime (λ_Leslie; R² = 0.523, P < 0.0001), indicating the need to adjust individual fitness for demonstrated changes in population growth and thus producing a relative individual fitness index (λ_rel). We regressed λ_rel on age at first reproduction, and found significant selection favoring earlier reproductive success (β ± SE = –0.20 ± 0.06; R² = 0.306, P < 0.0001). When using an earlier (offspring at weaning) versus later (those that survived their first hibernation) measure of fecundity, we found that the latter introduced considerable variation, likely environmental, into the estimate of selection. This greatly weakened the regression of relative fitness on the age at first successful reproduction. LRS and LRS_rel exhibited nonsignificant changes with age at first reproduction. Finally, those females that reproduced successfully at younger ages had similar litter sizes but significantly shorter life spans than females that matured when older, perhaps reflecting costs to early reproduction.