Distributed combined cooling, heating and power systems (DCCHP) are the main form of distributed energy systems. Currently, predicted values are used for source load data when calculating the DCCHP cost. However, the limitations of various load forecasting methods and renewable energy output forecasting methods can result in errors in the forecast data, which in turn affects the accuracy of the costing of the combined supply system. This paper establishes a comprehensive model of DCCHP with participating components, uses the Latin hypercube sampling technique to simulate fluctuations in the prediction errors of various types of load and renewable energy output forecasts, and uses the scenario reduction technique to extract typical scenarios that can cover most of the errors, and then performs cost calculations for each typical scenario to obtain the system cost with the prediction errors weighted by probability. Finally, simulations and sensitivity analysis are carried out for a typical DCCHP. The results show that the proposed method is able to calculate the actual cost of the DCCHP for different prediction values with different accuracy, providing an effective reference for practical system planning and design.