Serum samples (1,428) from 1,149 patients with chronic liver diseases and polytransfused subjects were tested for antibody to hepatitis C virus by first-generation enzyme immunoassays. Antibody to hepatitis C virus was detected in 87% of patients with transfusion-related chronic non-A, non-B hepatitis; 17.6% of patients with nonmalignant, chronic hepatitis B virus infection; 37.3% of patients with hepatocellular carcinoma; 14.3% of patients with alcoholic liver diseases; 22.2% of patients with cryptogenic cirrhosis; 76% of intravenous drug abusers; 16.4% of patients on hemodialysis; 1.8% of patients on peritoneal dialysis; 6.2% of kidney transplant recipients; and 3.1% of normal subjects. A high frequency of weakly positive results was found in "aged" samples: 61.9% of antibody to hepatitis C virus-positive patients whose sera had been stored for more than 2 yr had optical densities less than two times the cut-off values in contrast to 7.9% of those whose sera had been stored for less than 2 yr (p less than 0.0001). A significantly lower proportion of patients who had optical densities less than two times the cut-off values were reactive in subsequent samples, 27.5% vs. 87.5% (p less than 0.0001). On retests, only 70% and 56% of sera that were initially positive for antibody to hepatitis C virus remained antibody to hepatitis C virus positive using second-generation enzyme immunoassays and neutralization enzyme immunoassays, respectively. Our results suggest that retrospective studies on stored sera may have overestimated the prevalence of antibody to hepatitis C virus.(ABSTRACT TRUNCATED AT 250 WORDS)