A model of uncertainty reduction theory was tested that was derived from Berger and Calabrese's (1975) theory of initial interactions and recent extensions of the theory (e.g., Berger, 1979; Berger & Bradac, 1982) across three relationships (acquaintances, friends, and dates) in three cultures: Japan, Korea, and the United States. The model was tested using LISREL and found to be a reasonable fit to the data for all three relationships in all three cultures. The percentage of variance explained in attributional confidence was found to be lower for friends than for dates and acquaintances. For the United States data more variance was explained in attributional confidence in acquaintances than dates; however, the findings were reverse for the Japanese and Korean data. These findings were discussed in light of previous research on uncertainty reduction theory and relationship development. Shortcomings of the model also were pointed out based on early critiques of the theory (e.g., Bochner, 1978) and a recent elaboration that examined the context in which romantic relationships develop (Parks & Adelman, 1983).