BACKGROUND: Understanding the factors which determine a household's or individual's risk of malaria infection is important for targeting control interventions at all intensities of transmission. Malaria ecology in Tanzania appears to have reduced over recent years. This study investigated potential risk factors and clustering in face of changing infection dynamics. METHODS: Household survey data were collected in villages of rural Muheza district. Children aged between six months and thirteen years were tested for presence of malaria parasites using microscopy. A multivariable logistic regression model was constructed to identify significant risk factors for children. Geographical information systems combined with global positioning data and spatial scan statistic analysis were used to identify clusters of malaria. RESULTS: Using an insecticide-treated mosquito net of any type proved to be highly protective against malaria (OR 0.75, 95% CI 0.59-0.96). Children aged five to thirteen years were at higher risk of having malaria than those aged under five years (OR 1.71, 95% CI 1.01-2.91). The odds of malaria were less for females when compared to males (OR 0.62, 95% CI 0.39-0.98). Two spatial clusters of significantly increased malaria risk were identified in two out of five villages. CONCLUSIONS: This study provides evidence that recent declines in malaria transmission and prevalence may shift the age groups at risk of malaria infection to older children. Risk factor analysis provides support for universal coverage and targeting of long-lasting insecticide-treated nets (LLINs) to all age groups. Clustering of cases indicates heterogeneity of risk. Improved targeting of LLINs or additional supplementary control interventions to high risk clusters may improve outcomes and efficiency as malaria transmission continues to fall under intensified control.