The countries of southern Africa, although infected with HIV slightly later than those further north, nevertheless overtook eastern Africa's levels of prevalence during the mid 1990s and then experienced the world's most terrible epidemic. By 2004 the region had 2 per cent of the world's population and nearly 30 per cent of its HIV cases, with no evidence of overall decline in any national prevalence, which in several countries exceeded 30 per cent of the sexually active population. The chief issue in southern Africa is therefore to explain the speed and scale of epidemic growth. The obvious explanation is the region's history of white domination and the dramatic economic change and social inequality it had wrought. The view here is that this is true, but the connections were not always obvious, while, as everywhere in Africa, the scale of the epidemic was chiefly due to the long incubation period that enabled it to spread silently beyond hope of rapid suppression.

By chance, both the earliest definite indication of HIV in southern Africa and the best evidence of the silent epidemic anywhere in the continent come from the remote rural Karonga district of northern Malawi, bordering Tanzania and Zambia. Karonga's people, famed in colonial times for their education received from Scottish missionaries, had migrated as clerks and craftsmen throughout the industrial centres of southern Africa. This may first have exposed them to HIV. The virus's arrival in Karonga can be traced because the district experienced a mass campaign against leprosy and tuberculosis that included two total population surveys, in 1981-4 and 1987-9, each of which took and stored blood specimens from everyone in two sections of the district. All 44,150 specimens have been tested retrospectively for HIV, although only those from people aged 15-49 are included in the calculations. The results give a uniquely detailed picture of the dynamics of a local epidemic.

In the first round of investigation, none of the 1,041 specimens taken in 1981 had HIV. Four infected specimens were taken in 1982, one in 1983, and six in 1984, making a total of eleven in 12,979 specimens, or less than 0.1 per cent. Four were men and seven women.