Rancho Bernardo Study that in subjects with normal-to-impaired glucose tolerance, fasting insulin levels correlate positively with bone mineral density measured some 4 years later (1). They conclude that insulin is osteotrophic, presumably by the promotion of bone formation. Another epidemiological study, the Rotterdam Study, reported similar findings of a correlation between insulin levels and bone mineral density and proposed an anabolic effect of insulin on bone formation (2). Neither the Rancho Bernardo nor the Rotterdam groups in drawing inferences about bone formation from their data address the substantive body of literature on reduced bone formation in diabetes. From a 12-year longitudinal study of patients with diabetes, we have added to the considerable evidence that the decline of bone mineral density is attenuated in well-controlled type 1 diabetes and that bone mineral density is perhaps even augmented in type 2 diabetes (3). We found by bone histomorphometric indexes and osteocalcin levels that bone formation was low and that bone formation correlated negatively with long-term changes in bone mineral density. Based on our observations, we suggested a unifying hypothesis that explains the divergent findings of low bone density in patients with type 1 diabetes and of normal or increased bone density in patients with type 2 diabetes (3). Several possible mechanisms can be identified to explain low bone turnover in diabetes. Hypoparathyroidism is a condition in which bone formation is low and bone mineral density is often strikingly increased (4). In studies of diabetic osteopenia, McNair (5) described a" state of functional hypoparathyroidism" in type 1 diabetes. Short-term studies have correlated changes in parathyroid function with mitigation of bone loss by the administration of thiazide diuretics (6). Although higher insulin levels caused by thiazide diuretics could result in higher bone mineral density (1), decreased calciuria and higher serum calcium with decreased parathyroid hormone levels suggest another possible mechanism for the association of bone mineral density and longterm thiazide diuretic use. Perhaps alter-