Abstract Background Some religious dimensions have been associated with different health-related outcomes over many years. Attending religious services is one of these dimensions that were associated with hypertension, with inconsistent results. And religious involvement seems to be closely influenced by sociodemographic factors, such as education. Therefore, this study aimed to investigate the association between religious service attendance and hypertension according to levels of education. Methods We analyzed baseline data of the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Frequency of religious service attendance and presence of hypertension were assessed in all 15,105 participants at baseline. The analyses were stratified by two levels of education (less than high school and high school or more). Logistic regression models were used to obtain the association between religious service attendance and hypertension in both groups. Results For those with high school or more, attending religious services was positively associated with hypertension (adjusted odds ratio [OR] = 1.14, 95% confidence interval [CI] 1.02–1.28). In contrast, for those with less than high school, attending services was inversely associated with presence of hypertension (adjusted OR = 0.73, 95% CI 0.55–0.96). Conclusions There seems to be a paradox in the association of religious service attendance and hypertension depending on the level of education.