Plasma levels of dehydroepiandrosterone-sulfate (DHEA-S), dehydroepiandrosterone (DHEA), delta 4-androstenedione (delta 4), testosterone and 17 alpha-OH-progesterone (17-OH-P) were studied in 58 samples collected in 18 patients with congenital adrenal hyperplasia due to 21-hydroxylase deficiency, during long-term ambulatory treatment with hydrocortisone. At each visit the patients were classified as being either in good control (GC) or in poor control (PC), based on well-defined clinical, auxological and biochemical criteria. The results were analyzed in relation to the degree of control and to chronological age (CA), bone age (BA), body surface (BS) and pubertal development. The most clear distinction between the children with GC and those with PC is found for DHEA-S (p less than 0.001 for BA). The majority of the DHEA-S values in the children with GC are closely grouped and significantly below the normal limits for CA, BA, BS and pubertal stage (p less than 0.001). In contrast, the PC children have wide-spread values, most of them being within or above the normal limits. The difference between GC and PC is also significant for testosterone (p less than 0.01) and delta 4 (p less than 0.05), but not for DHEA. Of the five steroids studied, DHEA-S is the most specific, whereas testosterone is the most sensitive and especially useful in girls and in prepubertal boys. delta 4 and 17-OH-P are almost as sensitive as DHEA-S, but they are less specific. DHEA is the less valid criterium.