Specimens from a total of 162 patients were studied prospectively using a rapid nucleic acid synthesis inhibition assay and a plaque-reduction assay to detect interferon levels in serum and cerebrospinal fluid. The specificity, consistency, and diagnostic applications of the rapid assay were evaluated by intensively investigating each patient for possible viral illness. Forty-one percent of patients with proven acute virus infections (excluding those with acute hepatitis and infectious mononucleosis) had unequivocal detectable interferon in serum or cerebrospinal fluid, compared with only two patients with bacterial infections (7%) who had equivocal interferon in the serum. Retrospective analysis showed the test to have high specificity for virus infections, but in patients with clinical syndromes of uncertain aetiology the test was not a particularly sensitive indicator of the presence of subsequently detected virus. The presence of interferon was not related to the time during the illness at which the sample was taken, nor to the numbers of lymphocytes in the cerebrospinal fluid, nor to the patients' general symptoms.