There has been increased attention recently on models and modelling within the global science education field. Research has begun to skew towards a competence-based perspective of models and modelling, as teachers are experiencing challenges and do not have the required competence in modelling from either theoretical or practical perspectives. This study was designed to comparatively investigate pre-service science teachers' (PSTs) and in-service science teachers' (ISTs) modelling competence. A rating scale questionnaire was developed to assess meta-modelling knowledge. Additionally, a Black Box modelling task was designed to evaluate modelling practices and products by using two techniques: think-aloud and drawings. The resulting data was then coded and scored with validated rubrics. Quantitative analysis revealed that ISTs outperformed the PSTs in meta-modelling knowledge but they had an almost equal level in modelling practices and products, which were not at a satisfactory level. Furthermore, modelling practices and products were positively related, but no significant relationships were found between meta-modelling knowledge, modelling practices and products. Results of qualitative analyses further indicated higher-level practices were reflected in the analysis of correct model products, which was accompanied by sophisticated scientific knowledge and other advanced scientific skills. Implications of this study for science education research and teacher professional development are discussed.