Both steady-state AC and DC power flow models are commonly used for techno-economic studies of power systems. The DC-based approach limits the computational burden by assuming small-angle approximation, ignoring power losses, reactive power flows, and voltage variations. It, therefore, matters to understand if this approach affects system vulnerability, reliability, and contingency assessments. To this aim, we use the timesequential Monte Carlo simulation and an N-k′-1 scenario for reliability and contingency analyses, respectively. Further, we introduce a new index for vulnerability assessment. The IEEE reliability test system (RTS) and the modified RTS are modeled. The results show that the DC model underestimates the reliability indices by about 20% and more than 90% in a stressed network. We also show a small error of 5%, owing to the assumptions of the DC model, which can lead to inaccurate simulations concerning the cascading failures. Finally, the sources of the inaccuracy in the DC-based model are investigated. The results prove that AC power flow model should be privileged for the line capacity-based assessments.