A group-randomized, double-masked, phase III trial of a Streptococcus pneumoniae conjugate vaccine is being conducted in American Indian populations in the southwestern United States. Approximately 9000 infants will be enrolled in the primary efficacy cohort with vaccine allocation determined by community of residence. The trial is designed to continue until 48 cases of invasive pneumococcal disease due to vaccine serotypes have accumulated. Thirty-eight geographically and socially distinct areas were randomized within blocks formed by population size and geographic location. This design affords the opportunity to capture the effects of herd immunity (indirect effects) by estimating the impact of the vaccine intervention on nonimmunized infants. Group-randomized trials have challenging design and analysis features, many of which are discussed here in the context of the first such trial designed to lead to licensure of a drug or biologic in the United States.