Perimenopausal bone loss is considered to affect trabecular bone preferentially. Peripheral quantitative computed tomography (pQCT) quantifies trabecular bone mineral density (BMD) independently at the ultradistal radius. This article examines differences in pQCT BMD between late premenopausal and early postmenopausal women, comparing the differences with calcaneal ultrasound and axial dual energy X-ray absorptiometry measurements. One hundred nineteen normal perimenopausal women aged 45-55 yr who attended a randomized osteoporosis screening program were stratified by menopausal status into premenopausal (PRE: n = 79) and postmenopausal (POST: n = 40) groups. All measurements were lower in the postmenopausal group with the exception of ultrasonic velocity (PRE vs POST: 1397 +/- 53.8 vs 1421 +/- 58.5 m/s, p = 0.037). Total (391.8 +/- 52.9 vs 366.3 +/- 68.6 g/cm(3), p = 0.013) and subcortical (533.6 +/- 59.4 vs 504.3 +/- 79.8 g/cm(3) p = 0.018), but not trabecular (187.5 +/- 38.8 vs 173.2 +/- 46.6 g/cm(3), p = 0. 098) or cortical (561 +/- 53.4 vs 551.2 +/- 66 g/cm(3), p = 0.174), pQCT BMD measurements were significantly lower in the POST group, as were ultrasonic attenuation (79.4 +/- 16 vs 72.3 +/- 18.0 dB/Mz, p = 0.034), DXA spine (1.032 +/-16 vs 0.959 +/- 0.2 g/cm(2), p = 0.003), and all hip (p </= 0.001) measurements. Although body mass index (BMI) was positively and menopausal status and age negatively correlated with most bone mass measurements, adjusting for BMI did not alter the relative deficits in postmenopausal compared with premenopausal women. This study suggests that early postmenopausal bone loss at the radius preferentially affects subcortical, rather than trabecular, bone in the appendicular skeleton, which suggests preferential trabecular bone loss in the axial skeleton.