Eighty one women with vertebral osteoporosis were treated for up to 2 years with fluoride administered either as monofluorophosphate (MFP, 200 mg/day, i.e., 26.4 mg fluoride-ion) or sodium fluoride (NaF, 50 mg/day, i.e., 22.6 mg fluoride-ion). All patients received calcium supplementation (1 g of Ca2+/day) taken apart from NaF and in the same tablet for MFP. Despite almost similar fluoride dosage of both regimens, the early increase in the bone mineral density (BMD) of the lumbar spine was higher with MFP than with NaF, reaching 11% and 4%, respectively, at 1 year (p = 0.007), and 21% and 6%, respectively, at 18 months (p less than 0.001). The incidence of lower extremity pain syndrome related to benign stress microfractures was also higher with MFP than with NaF (35% and 15%, respectively, p less than 0.01). Urinary fluoride levels were higher in the MFP than in the NaF group (9.6 +/- 3.5 vs. 6.8 +/- 3.4 at one year, p = 0.003), suggesting that this difference in efficacy and tolerance is related to a better bioavailability of fluoride provided by MFP than by NaF. The occurrence of a stress microfracture could not be predicted by any clinical, biochemical, or densitometric parameter before treatment, but patients presenting with a stress microfracture during the course of the treatment had a higher gain in bone mass than those without stress fractures (at 1 yr+11 vs. +5%, p = 0.03 and at 18 months +18 vs. +6.9%, p less than 0.02). In conclusion, there is a clear correlation between the efficacy and the occurrence of side effects of fluoride therapy in osteoporosis.(ABSTRACT TRUNCATED AT 250 WORDS)