Summary As a proportion of AIDS-defining illnesses, Kaposi's sarcoma (KS) decreased from 1987-89 to 1993-94 in homosexual and bisexual men in all European regions and in the United States. Albeit underestimated, AIDS KS rates in the general male population at ages 25-49 are higher than those of the majority of cancer sites in the same age group.Keywords: Kaposi's sarcoma; AIDS; incidence A steady decline, in terms of the percentage of Kaposi's sarcoma (KS) as an AIDS-defining illness, was first reported among homosexual and bisexual men in the United States (US) (Des Jarlais et al, 1987). The broadening of the definition of an AIDS case in 1987 does not entirely explain this decline, as it persists when analyses of trends are restricted to KS and Pneumocystis carinii (i.e. two conditions that have always been part of AIDS definition) (Beral et al, 1990).The relative decline in KS was confirmed in AIDS surveillance data or clinical series in several developed countries, such as the United Kingdom (UK) (Peters et al, 1991), Germany (Schwartlander et al, 1992), Italy (Serraino et al, 1992) and Australia (Elford et al, 1993). It is at least partly attributed to the shorter latency period between HIV infection and KS onset than in other AIDS-defining illnesses (Hermans et al, 1996). According to some investigators (Elford et al, 1993;Dore et al, 1996), however, the relative decrease in KS might reflect a reduced prevalence and/or virulence of the postulated KS agent following the adoption of safer sexual practices by homosexual men.Up to the end of 1994, approximately 440 000 AIDS cases had been recorded in official data in the US and about 140 000 in Europe. About 20% of AIDS patients had KS at the time of presentation. It is therefore possible to review systematically recent trends in KS in major European regions, and to assess separately non-homosexual men and women, groups that were little studied in the early phase of the epidemic on account of the low frequency of the disease. Finally, in order to quantify the minimum size of the epidemic of AIDS-associated KS at a population level, age-standardized incidence rates have been computed from AIDS surveillance data.