The global HIV/AIDS epidemic has displayed heterogeneous geographical patterns at the various scales of analysis and perspective. Using the case of Kenya, the spatial patterns of the HIV/AIDS epidemic were examined at three levels, the regional, the local, and the individual level. HIV prevalence data on pregnant women from 1990 to 2004 from sentinel sites with continuous surveillance data since the early 1990s were aggregated at the provincial administrative level. Trend curves for each of Kenya’s eight provinces were derived, from which spatial-temporal maps using geographical information systems indicated an epidemic that had diffused in Kenya from provinces located in the west to the rest of the country. The prevalence trends over the surveillance period showed an epidemic that was spatially heterogeneous as the provinces exhibited different infection rates which reached the optimum levels at different times starting from as early as the early 1990s until 2000. Bivariate analysis of the provincial prevalence against socioeconomic and demographic characteristics of the country in 1990, when the HIV surveillance started, and in 2000, when the prevalence reached its highest level, showed that ethnic groups distribution as the strongest factor influencing the spatial patterns at the provincial level. Poverty and income inequality showed positive relationship with the provincial HIV prevalence rates, with the relationship more pronounced in provinces characterised by high income inequality. Migration and urbanization on the other hand showed only moderately low correlation with the HIV prevalence.