This study primarily aimed to estimate the association between influenza vaccination and the occurrence of hospitalization for acute respiratory or cardiovascular diseases, or all-cause death during the influenza season in an elderly population in South Africa. We conducted a nested case-control study using data from a cohort of 45 522 elderly members of a private medical funding organization during the moderate 2004 influenza season. In 1282 (2·8%) subjects the combined outcome occurred and the influenza vaccination rate in controls was 15·4%. After adjustments for measured confounders, vaccination was associated with a statistically significant reduction of 19% (95% confidence interval 3·1–32·9) in the combined outcome. Post-hoc sensitivity analysis of the potential impact of potential healthy user bias showed that confounding, if present, could have caused this finding. Our data were inconclusive regarding the benefits of influenza vaccination in elderly persons in South Africa and given the low vaccine uptake, long-term follow-up is warranted.