Tversky's (1977) diagnosticity principle implies that categorization affects similarity, and that similarity in turn is based on context. However, Nisbett, Peng, Choi, and Norenzayan (2001) suggest that Chinese and Westerners differ in their sensitivity to context and categorization. Because of these differences, it is not clear whether Chinese should follow the diagnosticity principle. To explore these possibilities, we conducted a cross-cultural experiment using participants from Australia and China to repeat the experiment of Tversky (1977) using schematic faces as stimuli. Results showed that Australians, but not Chinese, made similarity judgments in a manner compatible with the diagnosticity principle. We suggest that: 1) the use of the diagnosticity principle depends upon contextual variables for Chinese people; and 2) Chinese participants judged neutral schematic faces as more positive than Western participants did.