To determine whether serum 3 alpha-androstanediol glucuronide (3AG) reflects the overall effect of integrated adrenal androgen secretion in the virilizing form of congenital adrenal hyperplasia (CVAH), circadian levels (0800, 1200, 1600, and 2000 h) of serum 3AG and 17-hydroxyprogesterone (17OHP) or 11-deoxycortisol (S), androstenedione (A), testosterone (T), and 24-h urinary 17-ketosteroids (17KS) were examined in seven patients (pts) with classical 21-hydroxylase deficiency (21OHD) and one pt with classical 11 beta-hydroxylase deficiency (11 beta OHD). Hormonal studies were conducted during the second day of dexamethasone (Dex) administration (2 mg/day). In five poorly controlled CVAH pts, including the 11 beta OHD pt, highly elevated baseline morning (AM) serum 17OHP or S as well as A levels, and elevated AM T levels in three pts decreased markedly in the evening (PM), while elevated serum 3AG showed no significant circadian changes; 17KS levels were markedly elevated for age. During Dex, moderately or slightly elevated AM 17OHP, A, or T in two to four pts with 21OHD decreased to the normal range in the PM. In the pt with 11 beta OHD, S, A, and T levels were suppressed. 3AG levels were modestly elevated or normal, without circadian changes, in these pts; 17KS levels were elevated or normal. In two other 21OHD pts, modestly elevated AM baseline 17OHP and A levels decreased in the PM; elevated AM T decreased in one pt in the PM; modestly elevated 3AG levels showed no circadian changes; 17KS levels were modestly elevated. During Dex, normal or slightly elevated serum steroids and 17KS levels were associated with normal or high normal 3AG levels without circadian changes. In one postpubertal female with 21OHD, modestly elevated AM baseline 17OHP levels decreased at 2000 h; normal A and T levels throughout the day and low normal 17KS were associated with slightly low 3AG levels, without circadian variation. During Dex treatment, normal 17OHP, A, T, and low 17KS levels were associated with low 3AG levels without circadian variation. In all pts as a group, an excellent correlation (r = 0.9) was found between either 0800 h or mean, or 2000 h serum 3AG levels and 17KS. In addition, AM and PM serum 3AG levels in five normal women were similar. We conclude that the high correlation between serum 3AG and urinary 17KS and the absence of a significant circadian variation in 3AG indicate that serum 3AG, regardless of sample time, is a useful metabolic index of integrated adrenal androgen secretion in CVAH.