Overcooling causes thermal discomfort and wastes energy. Effectiveness of a novel dedicated outdoor air system (DOAS) consisting of a multi-stage direct expansion coil to generate extra-low temperature (XT) outdoor air to handle the entire space cooling demand for better energy and humidity control has been confirmed in previous studies. In these studies, the terminal system was assumed using variable air volume (VAV) without reheating. The conventional VAV system without reheating has overcooling problem. This is associated with uncertain fluctuations of operating characteristics. However, virtually no study has been done to investigate its overcooling risk, and how it is compared with XT-DOAS. In this study, a rigorous probabilistic approach is proposed to account for the uncertain fluctuations of operating characteristics to assess their overcooling risks. Monte Carlo simulations with 10,000 interactions was used for risk analysis. A morphing method was employed to develop 10,000 weather files that could give an accurate prediction of future variations in yearly weather conditions. Reliable datasets were used to develop the probability distribution functions to account for the daily variations in use patterns. The most probable operating characteristics were input to EnergyPlus for hour-by-hour simulations. The annual cooling load profile formulated by the probabilistic approach was successfully validated by in-situ measurements. The probabilistic assessment results show that XT-DOAS has far less overcooling hours and lower long-term percentage of dissatisfied than the conventional VAV system. The results also indicate that the proposed probabilistic approach is useful for a robust system performance evaluation.