In this article we describe the development of a highly sensitive, accurate, and reproducible RIA for the measurement of 3,3'-diiodothyronine (3,3'-T2) in human serum and brain tissue. The detection limits were 1.8 fmol/g and 1.5 pmol/L in human brain tissue and serum, respectively. Serum concentrations of 3,3'-T2 were measured in 4 groups of patients with nonthyroidal illnesses (NTI), i.e. brain injuries (n = 15), sepsis (n = 24), liver disease (n = 22), and brain tumors (n = 23). The mean serum concentration of 3,3'-T2 in 62 healthy controls was 46.6 +/- 20.0 pmol/L. 3,3'-T2 levels declined significantly with increasing age. They were significantly lower in patients with brain injury (34.2 +/- 19.4 pmol/L; P = 0.006), were at the upper limit of normal in patients with sepsis (57.0 +/- 36.9 pmol/L; P = 0.06), and were elevated in patients with liver disease (72.6 +/- 56.7 pmol/L; P = 0.04) and brain tumors (89.0 +/- 40.9 pmol/L; P = 0.01). The serum levels of T3 were significantly lower than those in controls in all 4 patient groups. Serum concentrations of 3,3'-T2 were significantly enhanced in 9 patients with hyperthyroidism (85.4 +/- 43.0 pmol/L; P = 0.01) and were reduced in 12 patients with hypothyroidism (14.9 +/- 9.2 pmol/L; P = 0.001). In both normal brain tissue, obtained either intraoperatively or excised postmortem, and brain tumors, the concentrations of 3,3'-T2 ranged between 50-300 fmol/g. In healthy controls, 2 different forms of acute stress (sleep deprivation and delivering a lecture) significantly increased serum levels of T4 and T3, but did not affect those of 3,3'-T2 or 3,5-T2. In conclusion, our results show that, contrary to expectation, a low T3 syndrome in NTI is not always associated with low serum concentrations of 3,3'-T2. The production of 3,3'-T2 in NTI seems to be regulated in a disease-specific manner, resulting in unchanged, reduced, or elevated hormone concentrations.