Abstract Background Many previous studies estimating the relationship between body mass index (BMI) and dental decay are conflicting. Most studies, however, examine the relationship using BMI as a categorical variable. This study evaluated the non-linear association between body mass index as a continuous variable and untreated dental decay. Methods Cross-sectional study of adults free of diseases attending a tertiary dental clinic was conducted. The number of untreated caries at the time of consultation was assessed using the WHO criteria. A multivariable Poisson regression model for severity of untreated dental decay was first established. Restricted cubic spline functions were used to consider potential non-linear associations between BMI and untreated dental caries. Results After multivariable adjustment, the prevalence ratios (PR) for the number of dental decay remained significantly associated with the age at beginning tooth brushing (PR = 1.15, 95% CI 1.05–1.25), BMI normal (PR = 1.30, 95% CI 1.03–1.65), SDI (PR = 0.61, 95% CI 0.50–0.75) and GI (PR = 1.59, 95% CI 1.30–1.94). When BMI was evaluated as continuous variable, it exhibited a significant U-shaped pattern with the number of untreated dental decay both in univariable and multivariable analysis. Conclusion The rate of untreated tooth decay was associated with both under- and overweight status.