Spontaneous fractures were reported to be rare (less than 1%) in 1664 hospital admissions for hip fracture in the 1950s in Sweden. We report 11 fluoride-treated postmenopausal patients who developed spontaneous fractures of the femoral necks, all subcapital initially. In 7 patients who continued treatment there were later femoral neck or shaft fractures; in 6, these were bilateral (one followed a fall). In all there were 19 spontaneous fractures: 5 were asymptomatic, including 2 with deformity; 12 fractures required surgery. Five were incomplete (stress) fractures. All were treated with supplementary calcium 1 g daily; 10 had vitamin D supplementation. In all patients where the timing was known, the initial and subsequent fractures were preceded by, or associated with increased bone turnover as measured by plasma alkaline phosphatase (pAlP) (i.e., they were all “good responders”). Two had pretreatment hip fractures following falls. We compared these 11 (Group 1) and another identically treated group of 14 patients (Group 2), without spontaneous femoral fractures and not different in mean age, pretreatment vertebral fractures, years since menopause, fluoride dosage, and plasma creatinine. Group 1 had a lower (p less than 0.05) index of cortical bone in the femoral neck, as assessed by the ratio “calcar width/femoral neck minimum width.” The 6 biopsied fluorotic patients from Group 1 had a higher (p less than 0.05) bone fluoride content than the 4 biopsied fluorotic patients from Group 2. Furthermore, histological cortical features of thinning, increased porosity, and advanced tunneling resorption characterized Group 1 posttreatment biopsies. There were no significant differences in peak pAlP responses in the two groups. Mild asymptomatic vitamin D excess may have been a contributing factor in three Group 1 patients. Two further treatment groups have been studied more recently by forearm single-photon absorptiometry (SPA) at two sites; a cyclic NaF group (Group 3) and a calcium +/- vitamin D group (Group 4). Neither showed significant changes in forearm cortical bone density on treatment for 2 and 1.5 years, respectively, but Group 3 showed a significant increase in density at an ultradistal (60% trabecular) site. The pAlP response in Group 3 was significantly less than in Group 1. Spontaneous femoral neck or shaft fractures did not occur in either Groups 3 or 4. Therefore, we recommend: (1) Avoidance of sodium fluoride (NaF) treatment if pretreatment femoral fracture or thin femoral neck cortices exist. (2) Reduction in NaF dosage if pAIP or osteocalcin increases by over 50% of pretreatment values. (3) Cessation of NaF if femoral neck or shaft fracture occurs during treatment, and also on the attainment of early to moderate spinal fluorosis. (4) Consideration of cyclic NaF treatment, to limit overstimulation of bone turnover and to enhance mineralization.