We developed and validated a RIA for measuring serum activin A. The least detectable value of this assay was 0.1 micrograms/L, and the antibody used cross-reacted slightly with bovine inhibin (3.2%) and porcine activin AB (10.0%) but not with porcine activin B (< 0.5%). Serum activin A was extracted with acetonitrile and trifluoroacetic acid to get rid of the interaction with possible binding proteins in serum. As a result of this extraction procedure, the dose-response curve of serum extract was parallel to the standard curve and a single immunoreactive (ir-) peak was demonstrated on gel chromatographic analysis with constant recovery rates over 80%. Serum ir-activin A level in healthy adults was 1.27 +/- 0.03 micrograms/L (mean +/- SEM, n = 180); being 1.38 +/- 0.05 micrograms/L (n = 90) in male, and 1.16 +/- 0.05 micrograms/L (n = 90) in female subjects, with a tendency to increase with age. Serum ir-activin A level during pregnancy showed a marked increase with the advance of gestation; 1.65 +/- 0.41 micrograms/L (n = 7) in the early, 4.50 +/- 1.13 micrograms/L (n = 21) in the middle, and 16.32 +/- 2.25 micrograms/L (n = 26) in the late trimester, with a rapid decline after delivery. On the other hand, serum ir-activin A level was elevated in patients with hyperthyroidism (1.91 +/- 0.37 micrograms/L, n = 31), liver cirrhosis (2.03 +/- 0.71 micrograms/L, n = 10), chronic renal failure (3.41 +/- 0.34 micrograms/L, n = 41), and advanced solid cancer (2.24 +/- 0.52 micrograms/L, n = 67). These findings indicate that serum ir-activin A level varies with physiological conditions such as aging and pregnancy, and that it may reflect the altered production and metabolism of activin A in certain diseased conditions.