OBJECTIVES: Guidelines for Health Technology Assessment (HTA) and Comparative Effectiveness Research (CER) largely focus on pharmaceuticals and only few explicitly consider other health care technologies. CER of medical devices (MD) faces some challenges that raise questions about how adequate current CER methods account for the specific features of MD and how well MD fit in the paradigm of drug HTA. Our aim was to identify challenges and gaps in methodology related to specific issues of MD. Our comprehensive framework for the evaluation of clinical effectiveness of MD includes recommendations for generation of primary data and analyzing and synthesizing data in systematic reviews of CER of MD. METHODS: We performed a targeted literature review for CER methods and specific features of MD. An electronic database search was combined with systematic screening of tables of content of selected journals in the fields of epidemiology, HTA, statistics, and evidence-based medicine, which have a strong focus on methods. Additionally, we screened the reference lists of the most relevant papers. RESULTS: More than 200 publications about the general evaluation of MD and about specific CER methods were included. The MD’s physical mechanism of action, the dynamic development and regulatory evidence requirements are the driving features that suggest the increased use of certain methods for the evidence generation, finding of information for HTA, data analysis and synthesis, and interpretation of results. Rather than following the paradigms of drug evaluation, MD resemble more the notion of complex interventions. Our methodological framework is compatible with the EUnetHTA core model and integrates existing recommendations for other complex interventions. The consideration of observational data, operator characteristics, active control trials, and decision-analytic modeling are of special importance, as well as the application of Bayesian methods. CONCLUSIONS: The assessment of the clinical effectiveness of MD does require specific, although not necessarily new methods.