Secondary failure of plasma glucose control following initial successful response to diet therapy may be due to dietary indiscretion, or to progression of the intrinsic diabetic condition. We report a 10-year prospective natural history study of 432 newly diagnosed diabetic patients aged 40-69 years undertaken to assess the effect of intensive dietary management, where patients were transferred to insulin, or oral hypoglycaemic therapy (tolbutamide, metformin) by predetermined criteria of weight and plasma glucose. Secondary failure to diet therapy occurred in 41 patients in years 2-4, 67 patients in years 5-7, and 51 patients in years 8-10; 173 patients remained on diet alone until death or the end of the study. Continuation on diet alone was associated with a lower ongoing fasting plasma glucose, greater beta-cell function assessed by an oral glucose tolerance test at 6 months, and increasing age. The rate of rise of fasting plasma glucose was inversely related to the duration of successful dietary therapy, but mean weight remained constant in all groups while on diet alone. The ongoing fall in beta-cell function assessed by HOMA modelling closely mirrored the progressive rise in fasting plasma glucose: there was no change in mean insulin sensitivity in any of the groups.