The effectiveness of the 5E model has been supported by research in schools in recent years, although its efficacy has rarely been assessed in the long term to avoid novelty effects and to consider the impact of the usual loss of fidelity that time entails. This study was designed to assess the long-term effects on students' conceptual learning as a consequence of an intervention that introduced the 5E model to their teachers five years earlier. Two questionnaires measuring the prevalence of alternative conceptions about the kinetic-molecular theory were performed over a five-year interval (pre-test in 2014 and post-test in 2019). 725 students participated, 371 in pre-test and 354 in post-test. In schools where the 5E model was introduced, statistically significant differences were observed in the percentage of correct answers between the pre- and post-test in both grades (p = 0.003 and 0.0028, respectively), with Cohen's d = 0.427 and 0.439, respectively, indicating a relevant effect size, while in control schools no statistically significant differences were reported in either grade. These results provide strong evidence regarding the long-term sustainability of interventions aiming to change teachers' practices in favor of active learning methods such as the 5E model and their positive effects on students' conceptual learning.