A significant myeloprotective effect of melatonin in mice treated with etoposide, cyclophosphamide or carboplatin has been reported. The present study was designed to evaluate if the same effect could be observed in patients receiving chemotherapy. Twenty previously untreated patients with inoperable lung cancer received two cycles of carboplatin (given at area under the curve 5 by the Calvert formula) on day 1 and etoposide (150 mg m(-2) i.v.) on days 1-3 every 4 weeks. Melatonin 40 mg or placebo (double-blind) was given orally in the evening for 21 consecutive days, starting 2 days before chemotherapy. Patients were randomized to receive melatonin either with the first or the second cycle. Complete blood cell count with differential was done three times per week for 3 weeks. The median age of the cohort was 60 years (range 42-69), 16 patients had non-small cell and four patients small-cell lung cancer, 12 stage III and eight stage IV disease. In a multivariate analysis including age, sex, diagnosis, stage, performance status, doses of carboplatin and etoposide, and concomitant treatment with melatonin or placebo, the haematological parameters--depth and duration of toxicity for haemoglobin, platelets and neutrophils (ANC)--were not significantly different between cycles with/without melatonin. The mean ANC nadir and the mean number of days with ANC < 0.5 x 10(9) l(-1) were 0.5 x 10(9) l(-1) and 2.5 days, respectively, with/without melatonin. We concluded that, in patients with lung cancer, melatonin given orally at a dose of 40 mg per day for 21 days in the evening, does not protect against the myelotoxic effect of carboplatin and etoposide.