The purpose of this survey was to quantify potential relationships between dmft (dmfs) scores at 3.5 years of age and sociodemographic variables, parental characteristics, nutritional characteristics, oral health practice, and fluoride application.Dental health services for children 1.5 to 4 years old are provided at a health center in Niigata City. The dental health program is based on a six-month recall system. The program is composed of dental examinations, obtaining information through a questionnaire on the above-mentioned variables, individual dental health education for mothers based on the information obtained by the questionnaire, and fluoride application for children who desire it.This survey was conducted longitudinally in the program on children from 1.5 to 3.5 years of age. The information was obtained by examinations of tooth surfaces and by questionnaires on the abovementioned variables.The subjects analyzed were 286 children who received dental examinations and dental health services more than four times from 1.5 to 3.5 years of age.Stepwise multiple regression analyses (step up method) were used to develop models which are useful in explaining dmft (dmfs) scores by various combinations of independent variables. Eight dependent variables were analysed. These variables were dmft scores, upper anterior dmft, molar dmft, first molar dmft, second molar dmft, buccal-lingual dmfs, proximal dmfs and pit-fissure dmfs. The 23 independent variables were adopted from the information accumulated by the questionnaires filled out for children up to 3 years of age and were each indicated by a constant value, mode, or mean in character with these variables.The results showed that the independent variables (multiple correlation coefficient R=0.46, p<0.01) associated with dmft scores at 3.5 years of age were ‘Region’, ‘Sweets and beverages’, ‘Method of infant feeding’, ‘Parents and guardians’, ‘Mother's DMFT score’, ‘Period of infant feeding’, ‘Frequency of snacks’, ‘Age at start of weaning’ and ‘Experience of fluoride application’ in order of the highest partial correlation coefficient.The independent variables associated with the seven other dependent variables, which were the dmf scores according to tooth groups, tooth types, and tooth surfaces, were different according to the dependent variables. If the independent variables were the same, the order of the partial correlation coefficients in which the independent variable was associated with the dependent variable was different.The independent variable ‘Region’ was highly associated with the dmf scores of all tooth types and all tooth surfaces. Children from the urban area had lower dmf scores than children from two rural areas.The variables in relation to infant feeding habits, ‘Method of infant feeding’ and ‘Period of infant feeding’, were concerned with dental caries in teeth which erupted during the period of infant feeding. The longer children were fed, the higher the dmft score was. The number of dental caries was higher in breast-fed children.In the variables concerning sweets and beverages, ‘Frequency’, ‘Sweets and beverages’ and ‘Sucrose intake from foods which have a long clearance time’ were associated with dental caries. These variables are known to have an influence on the cariogenicity of foods containing sugar. ‘Mother's DMFT score’ was associated with her child's dental caries in molars. It is supposed that the occulusal form is likely to be inherent.‘Frequency of fluoride application’ was related to the prevention of dental caries in first molars which had optimum oppotunities for fluoride application from the time of eruption and had relatively more caries on the proximal surfaces.