Primary and secondary unresponsiveness to meglumine has long been described in human visceral leishmaniasis. However, no studies have been performed to elucidate if these therapeutic failures were due to strain variability in meglumine sensitivity or were related to host factors. We have studied the in vitro sensitivity of 37 strains of Leishmania infantum isolated from 23 patients (11 human immunodeficiency virus-infected and 12 immunocompetent patients) with visceral leishmaniasis. Sensitivity tests were performed by infecting murine macrophages with Leishmania parasites and culturing them in medium containing different concentrations of meglumine. For each test we calculated a 50% effective dose (ED50) corresponding to the meglumine concentration at which 50% of the Leishmania parasites survived. In vitro results were strongly correlated to immediate clinical outcome. All strains requiring an ED50 of >70 microg/ml were related to therapeutic failures, whereas all strains requiring an ED50 of <40 microg/ml corresponded to an initial efficiency of meglumine. Among those patients who were initially improved, relapses occurred in all immunocompromised patients and in most immunocompetent patients who had a short duration of treatment (15 days). Finally, we found that in vitro sensitivity of strains decreased progressively in relapsing patients treated with meglumine. Consequently, the physician may be encouraged to alternate meglumine with other treatments such as amphotericin B or pentamidine, especially in the case of relapsing patients.