Based on the data reviewed, it is necessary to conclude that diabetes is associated with profound changes in HDL metabolism. However, once we go beyond this simple generalization, it is apparent that the relationship between diabetes and HDL metabolism is not a simple one. A good deal of the complication evolves from the fact that IDDM and NIDDM seem to affect HDL metabolism quite differently, with the only apparent similarity the fact that plasma HDL-cholesterol concentration can be low in untreated patients with either IDDM or NIDDM. Thus, in patients with IDDM the primary event seems to be related to the insulin-deficient state, which results in a decrease in HDL turnover rate and resultant decline in plasma HDL-cholesterol concentration. In contrast, HDL turnover appears to be accelerated, not reduced in patients with NIDDM, and the low plasma HDL-cholesterol concentration is a consequence of the increased turnover rate. In addition, patients with NIDDM are not absolutely insulin deficient, and available evidence suggests that the higher the plasma insulin level, the lower the plasma HDL-cholesterol concentration in these patients. The differences noted above in the effect of IDDM and NIDDM on HDL metabolism are of great interest, and, unfortunately, not very well understood. There is, however, one additional difference, which may be of paramount clinical importance. For reasons not totally clear, plasma HDL-cholesterol concentrations in patients with IDDM treated with insulin are not lower than normal, and even tend to be higher than these values in a nondiabetic population. Possibly as a result of this phenomenon, there is no evidence that changes in plasma HDL-cholesterol concentration play a role in the development of macrovascular complications in IDDM. Although it is apparent from the considerations discussed in this review that a great deal more needs to be learned about the effect of insulin deficiency on HDL metabolism, changes in HDL metabolism do not appear to be clinically important in patients with IDDM. Unfortunately, this does not appear to be the situation in patients with NIDDM. Plasma HDL-cholesterol concentrations are lower than normal in patients with NIDDM, and this finding seems to be related to increased morbidity and mortality from CAD. Furthermore, there is no form of anti-diabetic treatment, irrespective of how effective it has been in achieving glycemic control, that has been shown to substantially increase plasma HDL-cholesterol level. Indeed, it has been difficult to demonstrate a consistent effect of any therapeutic approach on plasma HDL-cholesterol concentration.(ABSTRACT TRUNCATED AT 400 WORDS)