OBJECTIVE: The World Health Organization (WHO) has recently introduced a new world standard population for the production of age-standardized rates. In this study we compare cancer rates standardized to this population with those computed using reference populations in current practice, particularly the world standard of Segi (1960), in order to evaluate their adequacy as estimators of relative risk in diverse population groups and over time. METHODS: Incidence and mortality rates standardized using these reference populations were calculated and compared for various cancers. Standardized rate ratios were compared with more efficient methods of approximating relative risk, the Mantel & Haenszel and maximum-likelihood estimators. The differences were tested by taking a synthesis of the relative risks and by taking into account whether effects were homogeneous across age strata or not. RESULTS: There were no statistically significant differences between the relative risk estimates based on direct standardization and those obtained using Mantel & Haenszel (p < 0.99), or maximum-likelihood techniques (p > 0.99), regardless of whether the Segi or the WHO world population was used as the standard. CONCLUSIONS: Ratios of rates age-standardized using the world standard of Segi approximate relative risk as precisely as the WHO standard. For this, and important practical reasons, it is considered unnecessary to replace the Segi standard population for comparisons between cancer rates.