We review studies on the efficacy against disease caused by Streptococcus pneumoniae of the 23-valent polysaccharide pneumococcal vaccine in adult populations in the more developed countries. Meta-analyses of primary vaccine trials have attempted to reduce uncertainty from lack of power. Vaccine efficacy calculated from studies in South African gold-miners and in Papua New Guinea, with high attack rates and differing serotype patterns, cannot automatically be applied to more developed countries. Meta-analyses will overestimate a protective effect if this clinical heterogeneity is ignored. Meta-analyses limited to trials in the more developed setting show no protective effect against pneumococcal pneumonia and a non-significant protective effect against bacteraemia. Lack of a specific diagnosis limits the ability to detect a protective effect against pneumococcal pneumonia. Most, but not all, observational studies confirm a protective effect against bacteraemia. An effect on mortality in more developed countries has yet to be documented.