OBJECTIVE: Zidovudine is a well known cause of macrocytosis. However, many HIV-infected patients develop macrocytosis even though they do not receive zidovudine. The aim of this case-control study was to evaluate other causes of high mean corpuscular volumes (MCV) in HIV infected patients. METHODS: Thirty patients with a MCV > or = 100 fl (cases) were compared to 60 randomly selected controls with MCV A< or = 99 fl, none of them receiving zidovudine. RESULTS: Sex ratio, age, mean CD4, prevalence of alcohol abuse and liver disease were similar in both groups. Vitamin B12 or folic acid levels were not decreased in patients with macrocytosis. In contrast, there was an association between macrocytosis and use of stavudine alone or in combination with another antiviral drug (28/30 cases, 15/60 controls, odds ratio 40.6 (95% confidence interval 5.1-325.2), P< 0.001). The haemoglobin concentration among cases (mean 13.5 g/dl) was similar to that of controls (mean 13.0 g/dl). CONCLUSIONS: Stavudine use increase the risk of macrocytosis more than 40-fold in HIV-infected patients who do not receive zidovudine. None of the stavudine-treated patients, however, developed anaemia or had to interrupt treatment because of this side effect.