In this study of fertility change in Thailand, some specific propositions from traditional demographic theory are tested. The multilevel model of contextual and individual-level effects on current fertility includes contextual variables on the status of women, economic role of children, infant mortality rate, and percentage of women single. Control variables are age and parity. Individual variables are educational status, husband's occupation, and migration status. The goal of the study was to measure the direct and indirect effects on rapid fertility decline from the late 1960s to the late 1970s in the initial phase of fertility transition. A discussion of the theories of fertility decline emphases social structural variables. Data were obtained from microdata samples from the 1970 and 1980 censuses of Thailand. The own children methodology developed by Cho was used to match surviving children to the mother in the household. The dependent variable was weighted to represent the number of children a woman would have throughout her reproductive career. In order to obtain an annual, synthetic cohort rate, the total number of surviving children was divided by 4 and then multiplied by the number of years in the appropriate age category of women. The index of current fertility was adjusted upward to account for the proportion of unmatched children in the sample. Surviving children represented births occurring during 1965-69 and 1975-79. Women are grouped at ages 15-29 and 30-44 years. Infant mortality was obtained from estimates generated by Knodel and Chamratrithirong and Pejaranonda. Descriptions are provided of the fertility change between 1965-69 and 1975-79 and changes in social structure. The analysis was begun with a microlevel model; an analysis was conducted on then the impact of contextual characteristics on the predictors (regression coefficients). Several models were generated based on regression standardization. The results showed that changes in macro level variables appear as key predictors, but in the cross-sectional analysis, the impact of macro level variables was not strong. Changes in the status of women and the economic roles of children accounted for about held of the variation: micro level variables were less important. The discrepancy between longitudinal and cross sectional results was discussed.