This paper presents effective methodologies for reliability assessment of electromagnetic devices and systems. To achieve the goal, the first-order reliability method and the moment method are adopted to deal with probabilistic constraints taking uncertainties of design variables into account. The proposed methods are applied to the TEAM Workshop Problem 22 and the results are compared to those of Monte Carlo simulation in terms of accuracy and efficiency. The robustness of three different designs (initial design, deterministic optimum and robust optimum) is assessed from the reliability point of view. Among them, it is shown that the robust optimum is the best design which has a high reliability to two probabilistic constraints considered.