BACKGROUND: In a cluster randomized trial (CRT) of a Vi polysaccharide vaccine against typhoid in the slums of Kolkata we found evidence of vaccine herd protection. However, transmission of typhoid into clusters from the outside likely occurred in this densely populated setting, which could have diminished our estimates of vaccine herd protection. METHODS: Eighty clusters (40 in each arm) were randomised to receive a single dose of either Vi or inactivated hepatitis A vaccine. We analysed protection for the entire cluster and for subclusters consisting of residents of the innermost households. RESULTS: During 2 y of follow-up, total protection was 61% (95% CI 41 to 75), overall protection was 57% (95% CI 37 to 71) and indirect protection was 44% (95% CI 2 to 69). Analyses of the innermost 75% and 50% of households of the clusters showed similar findings. However, in the innermost 25% of households of the clusters, total protection was 82% (95% CI 48 to 94) and overall protection was 66% (95% CI 27 to 84). There was not a sufficient sample size to demonstrate such a trend for indirect protection in these innermost households. CONCLUSIONS: The findings suggest that analyses of the entire cluster may have led to underestimation of herd protection against typhoid by Vi vaccine and that restriction of the analyses to the inner subclusters may have led to a more accurate estimation of vaccine herd effects.