We developed a model to quantify the effect of factors influencing the spatio-temporal distribution of foot-and-mouth disease (FMD) in Tanzania. The land area of Tanzania was divided into a regular grid of 20 km x 20 km cells and separate grids constructed for each of the 12-month periods between 2001 and 2006. For each year, a cell was classified as either FMD positive or negative dependent on an outbreak being recorded in any settlement within the cell boundaries. A Bayesian mixed-effects spatial model was developed to assess the association between the risk of FMD occurrence and distance to main roads, railway lines, wildlife parks, international borders and cattle density. Increases in the distance to main roads decreased the risk of FMD every year from 2001 to 2006 (ORs ranged from 0.43 to 0.97). Increases in the distance to railway lines and international borders were, in general, associated with a decreased risk of FMD (ORs ranged from 0.85 to 0.99). Increases in the distance from a national park decreased the risk of FMD in 2001 (OR 0.80; 95% CI 0.68-0.93) but had the opposite effect in 2004 (OR 1.06; 95% CI 1.01-1.12). Cattle population density was, in general, positively associated with the risk of FMD (ORs ranged from 1.01 to 1.30). The spatial distribution of high-risk areas was variable and corresponded to endemic (2001, 2002 and 2005) and epidemic (2003, 2004 and 2006) phases. Roads played a dominant role in both epidemiological situations; we hypothesize that roads are the main driver of FMD expansion in Tanzania. Our results suggest that FMD occurrence in Tanzania is more related to animal movement and human activity via communication networks than transboundary movements or contact with wildlife.