PURPOSE: Exercise is understood to exert positive effects on bone. However cancellous bone has not been shown to increase with exercise. Previous results of our 1-yr observational prospective study in ovulatory women related 20% of the change in cancellous spinal bone mineral density (BMD), measured by quantitative computed tomography (QCT), to luteal phase length (the time from ovulation to menstruation, LL). METHODS: The 66 women who documented exercise daily included normally active women (N = 23) and those who ran consistently or were increasing running in preparation for a marathon (N = 43). Exercise did not affect BMD change in the women as a whole. We re-evaluated those data to determine whether exercise-related effects on spinal cancellous BMD change in regularly cycling premenopausal women were related to ovulatory characteristics. The potential relationship of exercise to BMD change was reanalyzed by stratifying women into tertiles according to average LL documented by quantitative basal temperature analysis. RESULTS: Repeated-measures ANOVA indicated independent positive effects of both luteal length (P = 0.001) and activity (P = 0.041). The 11 runners with LL > 10.9 d had a nonsignificant 0.5% increase in lumbar BMD while the 15 who averaged short LL (