Increased serum concentrations of FFA, bilirubin, and carboxylmethyl-propyl-furanpropionic acid, accumulating in chronic renal failure in direct relationship with serum creatinine, have all been impli cated in the pathogenesis of the low T-3 syndrome during illness. Cytokines may also be involved in the sick euthyroid syndrome. In contrast to interleukin-1 (IL-1) and tumor necrosis factor-alpha, IL-6 is usually detectable in serum during illness and acts as a systemic hormone. We studied the association between serum T-3 and IL-6 in consecutive hospital admissions with a wide variety of medical conditions. Patients were divided into group A (T-3, greater than or equal to 1.30 nmol/L; T-4, greater than or equal to 75 nmol/L; n = 41), group B (T-3, <1.30 nmol/L; T-4, greater than or equal to 75 nmol/L; n = 46), and group C (T-3, <1.30 nmol/L; T-4, <75 nmol/L; n = 13). Serum IL-6 levels in groups C and B were higher than those in group A (median values 59, 39, and 9 U/mL, respectively; P < 0.01). Serum creatinine and bilirubin/albumin ratios were similar in the three groups, but the FFA/albumin ratio in group C was higher than in group A (P < 0.05). When all patients were analyzed together, serum T-3 was negatively correlated to serum IL-6 (r = -0.56; P(0.001), bilirubin/albumin ratio (r = -0.29; P = 0.004), and FFA/albumin ratio (r = -0.21; P = 0.03), but not with creatinine (r = -0.16; P = 0.11). Stepwise multiple regression resulted in the following equation: serum T-3 = 2.13 - 0.181n(IL-6)- 0.15ln(creatinine) - 0.094ln(bilirubin/albumin) (r = 0.61). The variability in serum T-3 was accounted for 28% by In(IL-6), 5% by In(creatinine), and 4% by In(bilirubin/albumin). FFA/albumin did not contribute in this respect. We conclude that the low T-3 syndrome in nonthyroidal illness is associated with high serum IL-6 levels. However, even when IL-6 is assumed to play a causative role, the variation of serum T-3 in NTI-patients remains largely unexplained.