Path analysis was applied to data obtained in social surveys in Kumamoto and Sapporo, Japan, and Gothenburg, Sweden, to cross-culturally compare the causal models that describe the multiple stratum relationships between road traffic noise annoyance and endogenous and exogenous variables. Path models can estimate not only the direct effect of a variable on annoyance but also the indirect effect of the variable via other variables. The exogenous variables were selected from housing, personal and environmental factors, and the endogenous variables were selected from various activity disturbances and related effects, based on the results of correlation coefficients between variables and discrimination by factor analysis. An a priori path model was constructed at the start of the analysis and the structure equations for the endogenous variables were formulated. The standardized partial regression coefficients are called path coefficients and show the strength of the linkage between variables. A revised path model was constructed by deleting insignificant paths. The characteristics of annoyance responses were as follows: (1) annoyance caused by exhaust has the strongest relation to noise annoyance and (2) structures of noise annoyance were different between Japan and Sweden and between housing types, probably owing to differences in lifestyle.