Purpose: The PRECEDE-PROCEED (P-P) model is a framework for designing health education/promotion programmes. While the P-P model has been used to design community-based oral health programmes, outcome assessments of the model are seldom reported. The aim of the present study was to explore whether pathways in the P-P model accurately reflect the current status of environmental/behavioural assessment in a Japanese community and to examine whether using the linear structural relations (LISREL) programme would improve the model. Materials and Methods: In a community health centre in Japan, a questionnaire with 29 items regarding oral health was distributed to 824 mothers with 3-year-old children. The items were compiled into eight variables that were tested using the LISREL programme. Goodness-of-fit index (GFI), adjusted goodness-of-fit index (AGFI) and root mean square error of approximation (RMSEA) were statistically analysed before and after P-P model modification. Results: The GFI, AGFI and RMSEA were 0.913, 0.776 and 0.161 before, and 0.975, 0.939 and 0.075 after P-P model modification, respectively. Indirect effects on the quality of life from the predisposing, reinforcing and enabling factors became stronger than those in the initial P-P model. The overall fit of the modified P-P model was significantly better than that of the P-P model without modification, which did not accurately reflect the status of environmental/behavioural assessment in the community. Conclusions: The modified P-P model reflected the current status of environmental/behavioural assessment in the community. Health education models should be tested rigorously to ensure that they fit the reality of people's behaviour. Parole chiave: oral health promotion, LISREL analysis, PRECEDE-PROCEED model