OBJECTIVES: Decision-analytic modeling has become a widespread method applied in Health Technology Assessments (HTA), but the extent to which modeling is used differs widely among international HTA institutions. The German Agency for Health Technology Assessment (DAHTA) states in its methodological guidelines that model calculations can be carried out if necessary and feasible. However, DAHTA does not provide any methodological guidance in this regard. Aim of this study is to quantify the current role of decision-analytic modeling in DAHTA-reports and to analyze the applied methods. METHODS: All 140 DAHTA-reports published between 1998 and May 2011 were screened for the specific development of new decision-analytic models. To assess the impact of these models on recommendations, all relevant reports were reviewed with respect to the health economic conclusion, modeling methods and further research needs. RESULTS: A total of 90 DAHTA-reports incorporate an economic assessment. Of these, ten reports develop a new specific decision-analytic model. About 30% of the reports without a model come to a general economic conclusion but only one report gives a clear recommendation without major limitations. About 20% of these reports explicitly state that the development of a model for the German setting may have helped to come to a clear conclusion. In contrast, all reports incorporating a model give an economic recommendation – two of these with limitations. The identified models differ with respect to the type of health economic evaluation (cost-effectiveness, cost-utility), model type (decision tree, Markov model, Monte Carlo simulation), time horizon (two weeks – life long), discount rate (3%, 5%), perspective (statutory health insurance, care provider, social), outcome parameters (generic, disease specific) and sensitivity analyses (one-way, multi-way, probabilistic). CONCLUSIONS: Incorporating decision-analytic models in German HTAs has the potential to increase the number of health economic recommendations, but only a fraction of reports developed a specific model so far.