OBJECTIVES:   Randomized controlled clinical trials are the gold standard for determining causal inference.  However, trials are expensive, and the results can be difficult to interpret.  Our objective was to evaluate methods for clinical trial simulation to understand how the simulation approach can be used for improved trial planning and interpretation of trial results.  Our primary focus was trials of type 2 diabetes and cardiovascular disease.  METHODS:   We systematically searched the MEDLINE database for clinical trial simulation studies.  We used the MeSH terms: Markov model, Markov chains, simulation, simulation model, microsimulation, computer model, and required type 2 diabetes mellitus and cardiovascular diseases.  We restricted the search to studies of humans published in English and found 92 publications.  We also considered innovative clinical trial simulation methods from other areas to gain context.  RESULTS:   A number of established techniques — notably, the Archimedes Model, Markov models, and observational analyses— are used for clinical trial simulation.  Markov model-based simulations are widely employed, but have structural limitations with regard to the physiological detail they can capture (e.g. multiple comorbidities). Retrospective, observational methods for clinical trial simulation are gaining utility as more databases become available.  However, observational methods remain vulnerable to unknown biases.  Finally, large-scale simulation models (such as the Archimedes model), with physiological underpinnings, provide accurate and clinically detailed trial simulations.  These models are used to simulate trials of therapies not yet marketed, or to forecast late stage trials.  Model-based simulations require validations to ensure accuracy.  CONCLUSIONS:   Clinical trial simulation is an increasingly powerful tool, complementing real-world clinical trials.  Large scale simulation modeling has been shown to be valuable for estimating and interpreting clinical findings.  Recent studies suggest that future developments will leverage both large-scale simulation models and increasingly rich real-world evidence.