The uncertainties associated with the prediction of long-term structural performance of existing reinforced concrete (RC) structures can be reduced by updating the model parameters using observational information. However, those parameters have been updated ignoring the spatial steel corrosion, and this assumption could lead to an overestimation of the structural reliability. In this paper, an approach is proposed to update the reliability of deteriorating RC structures considering the effect of the variables that cause non-uniform steel corrosion. The stochastic interpolation process of the particle filter method is utilized to update the model parameters based on the observational data obtained at discrete locations on the structures. In an illustrative example, the life-cycle reliability of a RC girder is presented. Visual inspection results of corrosion crack width are applied to update the reliability, and the effect of the observational information on the reliability updating of the RC girder is discussed.