Over the next several decades, Medicare costs are going to increase significantly. Previous analysis has looked at the potential range of costs using a deterministic, scenario-based analysis. This provides a range of potential outcomes but provides no information as to the likelihood of these outcomes. The analysis here systematically considers the uncertainty inherent in forecasting Medicare costs using a Monte Carlo approach that reflects both the uncertainty in forecasting health costs and the uncertainty association with economic and demographic projections. Within 100 years, the 90 percent confidence interval for Medicare costs as a percent of GDP spans more than 30 percentage points.