Athletes have higher bone mineral density (BMD) relative to nonathletes. In amenorrheic athletes BMD may be compromised by estrogen deficiency, but it is unknown whether this is accompanied by structural differences. We compared femoral neck bone geometry and density of a-/oligomenorrheic athletes (AAs), eumenorrheic athletes (EAs), and eumenorrheic controls (ECs). We recruited 156 women: (68 endurance athletes and 88 controls). Femoral neck BMD, section modulus (Z), and width were measured using dual-energy X-ray absorptiometry. Menstrual function was assessed by questionnaire and classified as EA (≥10 periods/year) or AA (≤9 periods/year): 24 athletes were AA and 44 EA. Femoral neck BMD was significantly higher in EA than AA (8 %, difference) and EC (11 % difference): mean [SE] 1.118 [0.015], 1.023 [0.020] and 0.999 [0.014] g cm(-2), respectively; p < 0.001. Z was significantly higher in EA than EC (11 % difference): EA 667 [19], AA 625 [21], and EC 592 [10] cm(3); p < 0.001. Femoral neck width did not differ between groups. All differences persisted after adjustment for height, age, and body mass. The higher femoral neck Z and BMD in athletes, despite similar width, may indicate that exercise-related bone gains are endosteal rather than periosteal. Athletes with amenorrhea had smaller increments in bone mass rather than structural adaptation. The maintained femoral neck width in controls may be an adaptive mechanism to conserve bone strength in bending despite inactivity-related bone decrement.