Forty-eight female patients with postmenopausal osteoporotic vertebral compression fractures were treated with sodium fluoride and calcium supplements; their response to treatment was documented by sequential measurements of vertebral and forearm bone mineral density (BMD). During treatment 25 patients developed significant side-effects due to fluoride, and ultimately, 18 patients (37%) were intolerant of the drug after 17.3 +/- 7.3 (+/- SD) months. The remaining patients were followed for 29.4 +/- 9.6 months. By linear regression analysis, 69% of patients had a positive slope of vertebral BMD vs. time of greater than 0.0017 g/cm2.month (range, 0.0017-0.01) and were classified as treatment responders. The increment in vertebral BMD above the baseline value over time was described by the relationship delta BMD (g/cm2.month) = 0.042 + 0.0053 x months, equivalent to a rate of 8.4%/yr. Because the rates of increasing BMD were very variable, it was difficult to determine in individual patients the point at which a positive response to treatment had occurred; by calculating the 95% confidence target BMD by which the BMD must rise above the initial value before discounting the imprecision of the measurements (0.062 g/cm2), only 70% of responders were identified by 12 months. The total cumulative dose of sodium fluoride (31.3 +/- 16.4 g) was significantly higher in patients classified as responders than in the nonresponders (20.6 +/- 13.4 g; P less than 0.05), probably because of differences in side-effects between the two groups. In contrast, forearm BMD fell significantly in the responders by an average of 7.7%/yr, suggesting the possibility of a preferential improvement in axial bone density at the expense of cortical bone. Thus, the majority of patients treated with fluoride respond with increasing vertebral BMD, but it may take 12-24 months to identify these individually.