Different hydrogen peroxide (H2O2)-scavenging mechanisms, and the conditions under which they operate, have been examined in promastigotes and amastigotes of Leishmania donovani. For promastigotes, the ability of the parasite to remove H2O2 was completely ablated by sonication whereas for sonicated amastigotes substantial loss of H2O2 from the phagocyte-fre¸e test system still occurred. In direct contrast, the ability of amastigotes, but not promastigotes, to remove H2O2 was markedly inhibited by aminotriazole or sodium azide. This suggested a role for haem-containing enzymes, catalase or peroxidases, as a protective H2O2-scavenging mechanism and was consistent with detection of catalase in amastigotes but not promastigotes using a spectrophotometric assay. Both forms of the parasite did, however, show reduced ability to remove H2O2 at 5–7°C indicating that additional enzymatic scavenging mechanisms may operate. Glutathione peroxidase activity was undetectable in either form of the parasite. The total thiol sink, glutathione (GSH) plus protein thiols, was greater in promastigotes but the ability to regenerate GSH via glutathione reductase was equivalent for promastigotes and amastigotes. Less temperature-dependent non-enzymatic mechanisms (e.g. an unsaturated lipid sink) also appear to contribute to removal of H2O2 by both promastigotes and amastigotes. It seems likely, nevertheless, that the difference in H2O2 sensitivity between the two forms of the parasite relates to the activity of the direct H2O2-scavenging enzyme, catalase, which appears to operate more efficiently against a bolus of reagent H2O2.