To what extent do countries and people within countries vary in their perceptions about the risks and benefits of engaging in risky activities? And to what extent are these variations influenced by risk domains? We used generalizability theory to identify and quantify the variance components in perceived risk and benefit ratings that are attributable to differences between cultures, across raters within cultures, and across risk domains. Our results, based on two archival datasets from the United States and Japan, revealed that, although there is nontrivial variability in risk and benefit perceptions between American and Japanese raters, the amount of variability was smaller between cultures compared with across raters within cultures. Moreover, investment risk perceptions and gambling benefit perceptions were predominantly influenced by differences across raters within cultures, and hence, large numbers of raters would be needed to properly represent ratings of the corresponding populations. Our findings suggest that variability in risk and benefit judgments across raters from different cultures should not be simply attributed to pure cultural differences. Rather, a multi‐faceted approach, where multiple sources of variance are simultaneously considered, should be adopted to properly understand and effectively influence judgments of risk and benefit.