Background: This study aimed to investigate the association between metabolic syndrome (MetS) and periodontitis among young adults, and also to compare results using observed and latent variables for MetS and periodontitis. Methods: Data from the 1982 Pelotas Birth Cohort, Brazil, were used. Metabolic syndrome at the age of 23 years was measured using clinical and biochemical analysis and set as the main exposure. Periodontitis at the age of 31 years was clinically measured and set as the outcome. Confounding variables included sex and maternal education, assessed at birth, family income at 23 years, and smoking status at the age of 23 and 30 years. Factor analyses (exploratory and confirmatory) were performed to define latent variables for MetS and periodontitis. In addition, both conditions were also defined as categorical observed variables. The association between MetS and periodontitis was tested in structural equation models. Results: Two latent periodontal variables were identified: “initial” and “advanced” periodontitis, while one latent variable was identified for MetS. Metabolic syndrome is positively associated with “advanced” (coefficient 0.11; P value < 0.01), but not with “initial” (coefficient -0.01; P value = 0.79) periodontitis. When MetS and periodontitis were set as observed variables in the structural equation models, no association was found irrespective of the criteria used for periodontitis classification. Conclusions: There was a positive association between metabolic syndrome and “advanced” periodontitis, when the multiple dimensions of both diseases were accounted in latent variables. Nevertheless, when MetS and periodontitis were treated as observed variables, no association was detected irrespective of the criteria used for periodontitis classification.