Objective: This project was designed to longitudinally study persons who had antibodies to hepatitis C virus (HCV) to characterise the serologic course of infection. Methods: The subjects were 149 multitransfused patients (141 with thalassaemia major, 3 with thalassaemia intermedia, and 5 with sickle cell anaemia) who had been regularly followed up for 3 to 7 years. Sequential serum samples obtained semi-annually between January 1994 and January 2001 were tested, prospectively, by second or third generation HCV enzyme-linked immunosorbent assay (ELISA), followed by confirmatory recombinant immunoblot assay (RIBA-2 or RIBA-3). Results: Of the 149 patients, 90 did not seroconvert to HCV, whereas 59 had detectable antibodies. On the basis of RIBA results in these 59 patients, 24 (41%) had persistent high antibody levels to structural and non structural HCV antigens, 11 (19%) had persistent low antibody levels, 17 (29%) showed fluctuating antibody levels, and in 5 patients (8%) there was a total or a partial disappearance of specific antibodies (seroreversion), mainly anti-core antibodies. Two patients (3%) had antibody responses that did not fit into any of these four categories. In patients with fluctuating antibody levels, there were periods ranging from 6 months to 2 years when anti-HCV antibodies could not be detected. Conclusion: This study shows that the antibody response to HCV in patients who receive frequent blood transfusions is very variable. Individuals who exhibit intermittent seropositivity are a challenge to diagnosis.