Objective: To assess the effect of interventions on the levels of physical activity of healthy adults, users of the Brazilian Unified Health System and attended by the Family Health Strategy. Methods: Non-randomized experimental study with 157 adults allocated in three groups: 1) physical exercise classes (n = 54), 2) health education (n = 54), 3) control (n = 49). The study lasted for18 months, with 12 months of interventions and six months of follow-up after intervention. Assessments took place at the beginning, in the 12 months, and in the 18 months of study. Physical activity has been assessed by questionnaires and accelerometry. For the analyses, we have used the intention-to-treat principle and generalized estimating equations. Results: After 12 months, both intervention groups have increased the minutes of weekly leisure time physical activity and annual scores of physical exercise, leisure and transport-related physical activity. The exercise class group has obtained the highest average annual physical exercises score when compared to the other groups (p < 0.001). In the follow-up period, the exercise class group reduced its annual score (average: -0.3; 95%CI -0.5 to -0.1), while the health education group increased this score (average: 0.2; 95%CI 0.1 to 0.4). There have been no differences in the levels of physical activity measured by accelerometry. Conclusions: The interventions have been effective in increasing the practice of physical activity. However, we have observed that the health education intervention was more effective for maintaining the practice of physical activity in the period after intervention. We recommend the use of both interventions to promote physical activity in the Brazilian Unified Health System, according to the local reality of professionals, facilities, and team objectives.