1. Chronic marginal riboflavin deficiency was induced in groups of weanling rats by feeding a deficient diet supplemented with 0, 0.5, 1.0 and 1.5 mg riboflavin/kg diet. Ad lib.- and pair-fed controls received 3.0 and 15 mg riboflavin/kg diet respectively. 2. Serial measurement of erythrocyte NAD(P)H2 glutathione oxidoreductase (glutathione reductase; EC 1.6.4.2) and its activation coefficient revealed that after 12 weeks a steady-state of deficiency had been reached following initial fluctuations in status; the animals were then killed, and their tissues analysed. 3. Food intake, growth rate and the appearance of pathological signs were directly proportional to riboflavin content; however relative liver weight was increased above control levels only in the most-severely-deficient group, and anaemia was not detected in any group. 4. The activation coefficient of glutathione reductase in erythrocytes and liver was closely related to dietary riboflavin content; that of skin responded maximally even in the least-severely-depleted animals. 5. Hepatic and renal flavin contents were directly proportional to dietary riboflavin, FAD being conserved at the expense of riboflavin and FMN. ATP:riboflavin 5-phosphotransferase (flavokinase; EC 2.7.1.26) activity was reduced, even in the least-severely-deficient animals; ATP:FMN adenylyltransferase(FAD pyrophosphorylase; EC 2.7.7.2) was increased in liver, but only in the most-severely-deficient animals. 6. Hepatic succinate:(acceptor) oxidoreductase (succinate dehydrogenase; EC 1.3.99.1) activity fell sharply between 1.5 and 0.5 mg riboflavin/kg diet, producing an S-shaped dose-response curve; it showed smaller or less specific changes in other tissues such as brain, skin and intestine. NADH:(acceptor) oxidoreductase (NADH dehydrogenase; EC 1.6.99.3) activity declined in liver and intestine, but not in skin or brain. 7. The activation coefficient of glutathione reductase was correlated strongly with nearly all the riboflavin-sensitive variables measured, once equilibrium had been reached in this chronic deficiency model, and it was particularly strongly correlated with hepatic and renal FAD levels. Under equilibrium conditions, therefore, it appears to represent a good index of the extent of riboflavin deficiency, and significant changes in flavin levels and enzymes in the internal organs were detected even under conditions of marginal deficiency, associated with relatively small increases in the activation coefficient.