In the design of steel crane girders, various sources of uncertainty such as material properties, loads, and geometric tolerances are inherent and inevitable. Using deterministic structural and/or load conditions may lead to low-reliability systems in real applications. In this paper, the probability of failure of overhead crane bridge girders with uncertain design parameters is investigated. First, the design problem of a crane double girder is introduced within a set of analytical stress and defection constraints. Then, the response surface method is used in conjunction with Monte Carlo methods to quantify the effect of the parameter uncertainties on the constraints of stress and deflection. For illustrative examples, various configurations of girders with original deterministic parameters proposed in the literature are selected and their deterministic optimization values are considered as the mean of random variables. The obtained results reveal that uncertainties such as coefficients of variation (COV) in structures and loads have strong effects on the probability of failure for all stated crane girder configurations. For only a wheel load COV of 0.05 and geometric dimension COV of 0.025, the means of geometric parameters have to be larger than 1.1 their deterministic-based values in order to reach a probability of failure at a level of 10-4. Keywords: failure probability; double-box girder; overhead crane; Monte Carlo method.