According to the characteristics of urban road intersection, a probability index, namely reliability, was introduced to study the analysis and optimizing design theory of urban road intersection based on reliability analysis theory; meanwhile, a optimizing design model of urban intersection was proposed with the Numerical-approaching method, which could transform probabilistic constraint to deterministic constraint. The comparison of the proposed method versus a traditional Monte Carlo method was applied to a case study of X-urban intersection. The results show that the proposed model makes the calculation more efficient and easy. With the continued expansion of the modern city and the amount of car ownership, the amount of urban internal transport is also increasing dramatically. As a result, environmental pollution and traffic accident caused by traffic congestion are growing rapidly. Therefore, it is necessary to propose new requirements for urban road management. Current evaluation methods of road traffic running are mainly limited to static parameters, e.g., degree of saturation, congestion level, and occupancy, which are non-probability measure parameters. However, transportation system is a large dynamic system and the actual road capacity and traffic demand are constantly changing, due to various factors. It is necessary that establishing a probabilistic assessment index to evaluate the urban road system state. This paper presents the urban road intersection optimizing theory based on reliability analysis to solve traffic system optimization problems. Reliability is defined as the probability of required function or mission under required conditions. As a probabilistic index, reliability can be more accurate and comprehensive to evaluate the situation of urban road intersections and the stability of load system. Moreover, the operational quality of transportation system can be inspected by the index, as the basis for urban intersection management and design. It is more effective and practical than the deterministic index used in the past.