Seventy-nine osteoporotic (prior forearm or vertebral fracture), but otherwise healthy, postmenopausal women (aged 55 to 75 years) were allocated to two double-blind trials: (1) 39 women received either nandrolone decanoate (anabolic steroid) 50 mg as an intramuscular depot injection or a placebo injection every 3 weeks for 1 year; and (2) 40 women received either 2 mg 17 beta-estradiol plus 1 mg norethisterone acetate or placebo tablets daily for 1 year. Sixty-seven (85%) completed the 1 year of treatment. Serum concentration of type I procollagen carboxy-terminal propeptide (PICP) was measured before and at 3, 6, 9, and 12 months of therapy. In addition, 32 of the women had an iliac bone biopsy taken after double tetracycline labeling. Initial serum PICP correlated significantly with histomorphometrically measured rate of bone formation (r = .4; P less than .05) and plasma bone Gla protein (r = .6; P less than .001), but not with histomorphometrically measured bone resorption or biochemical estimates of bone resorption (fasting urinary hydroxyproline and calcium). Estrogen-progestogen therapy significantly decreased (P less than .001) serum PICP by about 30%, whereas anabolic steroid therapy hardly affected it. We conclude that serum PICP may be used as a noninvasive measurement of bone formation on a group basis. Whereas bone formation is clearly decreased during estrogen-progestogen therapy, it is not affected by long-term therapy with anabolic steroids.