We performed this study to access the changes in glucose tolerance over time in a group of women with polycystic ovary syndrome (PCOS) (n = 71) and control women (n = 23) with regular menstrual cycles and baseline normal glucose tolerance. Mean follow-up was between 2 and 3 yr for both groups (PCOS 2.5 +/- 1.7 yr; controls 2.9 +/- 2.1 yr). Based on World Health Organization glucose tolerance categories, there was no significant difference in the prevalence of glucose intolerance at follow-up in the PCOS group. In the PCOS group, 25 (37%) had impaired glucose tolerance (IGT) and seven (10%) had type 2 diabetes mellitus at baseline, compared with 30 (45%) and 10 (15%), respectively, at follow-up. There were also no differences within groups (PCOS or control) or between groups (PCOS vs. control) in the oral glucose tolerance test-derived measure of insulin sensitivity, but in the women with PCOS who converted to either IGT or type 2 diabetes mellitus, there was a significant decrease (P < 0.0001). At the follow-up visit, the mean glycohemoglobin level was 6.1 +/- 0.9% in women with PCOS vs. 5.3 +/- 0.7% in the control women (P < 0.001). Women with PCOS and baseline IGT had a low conversion risk of 6% to type 2 diabetes over approximately 3 yr, or 2% per year. The effect of PCOS, given normal glucose tolerance (NGT) at baseline, is more pronounced with 16% conversion to IGT per year. Our study supports that women with PCOS (especially with NGT) should be periodically rescreened for diabetes due to worsening glucose intolerance over time, but this interval may be over several years and not annually.