Background: We hypothesized that serum γ-glutamyltransferase (GGT) would positively predict the risk of microalbuminuria, a frequent consequence of both diabetes and hypertension, because serum GGT predicted diabetes and hypertension in dose-response relationships. Methods: In this prospective study, 2478 black and white men and women without microalbuminuria at year 10 provided urine samples 5 years later. Year 10 GGT cutpoints were 12, 18, and 29 U/L. Results: The incidence of microalbuminuria across year 10 GGT categories was U-shaped. Adjusted odds ratios across quartiles of serum GGT were 1.0, 0.39, 0.54, and 0.94 (P <0.01 for quadratic term), but the shape of association depended on the status of hypertension or diabetes (P <0.01 for interaction). Among individuals who ever had hypertension or diabetes, year 10 serum GGT showed a clear positive dose-response association with incident microalbuminuria (P <0.01 for trend), whereas among individuals with neither hypertension nor diabetes during the study, year 10 GGT showed a U-shaped association with it (P = 0.01 for quadratic term). When the long-term risk was evaluated in 3895 participants based on serum GGT at year 0 and prevalence of microalbuminuria at year 10 or year 15, the trends were similar but weaker than those of short-term incidence risk. Conclusions: Serum GGT within the physiologic range predicted microalbuminuria among patients with hypertension or diabetes and may act as a predictor of microvascular and/or renal complications in these vulnerable groups. GGT showed a U-shaped association with microalbuminuria among persons who did not develop either hypertension or diabetes.