This research examines ontological patterns of change in variation of the human pelvis as a means of identifying regions of differential growth, growth canalization and evidence of selection. Data were derived from pelvic radiogrammetry of 180 8-year-olds and 89 subjects at age 18 who were part of the Fels Longitudinal Growth Study. Coefficients of variation (CVs) and total growth increments were compared between sexes and between ages 8 and 18 for 14 pelvic measures. Sex-specific comparisons of mean size were tested per age using Student's t, whereas coefficients of variation were calculated and compared using the methods suggested by Sokal and Braumann ([1980] Syst. Zool. 29:50-66). The Mann-Whitney U-test was used to test median growth increments between ages 8 and 18. Results of these comparisons show significant sex differences in breadth of the ischium and acetabular regions among 8-year-olds. Most of the sexual dimorphism in the pelvis at age 18, however, develops during the adolescent growth period, during which both male and female pelves undergo growth remodeling of the pelvic cavity. Over the same time period, males show significantly greater incremental growth in the acetabulum, and females show differentially greater growth in the pelvic cavity. At age 18, the pelvis demonstrates a posterior-to-anterior gradient of increasing dimorphism within the inlet and midplane of the pelvic birth canal. As a means of interpreting the effects of natural selection on the pelvis, it is argued that appropriate comparisons are within-sample comparisons of CVs over time, rather than comparisons between sexes of adult coefficients as has been argued by others (Meindl et al. [1985] Am. J. Phys. Anthropol. 68:79-85). Analyses of change in coefficients of variation over time show evidence of concordantly reduced within-sample variation in 7 pelvic dimensions indicating canalization of growth. These results are attributed to the effects of stabilizing selection operating on both males and females and include transverse diameters of the sacrum, inlet, anterior inferior iliac spines, and breadths of the ilium and ischium. Six pelvic dimensions show evidence of increased total sample CVs and discordant change in within-sex comparisons of CVs as well as differential growth between sexes over time. This pattern is indicative of the effects of disruptive selection on the pelvis for interacetabular diameter, breadths of the anterior superior and posterior inferior iliac spines, public length, and ilium height.