Treatment of Human African Trypanosomiasis (HAT or sleeping sickness) relies on a few drugs which are old, toxic and expensive. The most important drug for the treatment of second stage infection is melarsoprol. During the last 50 years treatment failures with melarsoprol were not a major problem in Trypanosoma brucei gambiense patients. Commonly a relapse rate of 5-8% was reported, but in recent years it has increased dramatically in some important foci of T. b. gambiense sleeping sickness. Treatment failures for T. b. rhodesiense are much less of a problem apart from some reports between 1960 and 1985 of refractoriness in T. b. rhodesiense patients in East Africa. Analysis of those isolates revealed that their in vitro sensitivity to melarsoprol was one-tenth that of sensitive isolates, and complete failure to cure the infection in the acute mouse model with melarsoprol levels comparable with those in human patients. There was very little indication of resistance in T. b. gambiense isolates from Côte d'Ivoire and NW Uganda. The in vitro melarsoprol sensitivities for populations from relapsing and from curable patients were in the same range. Melarsoprol concentrations in the plasma and cerebrospinal fluid of patients 24 h after treatment did not show any difference between patients who relapsed and those who could be cured. The reason for relapses in the recent T. b. gambiense epidemics are not known. Other parasite-related factors might be involved, e.g. affinity to extravascular sites other than the CNS which are less accessible to the drug. In conclusion, a combination of factors rather than a single one may be responsible for the phenomenon of melarsoprol treatment failures in T. b. gambiense patients.