The intersection sight distance (ISD) design presented by AASHTO is based on extreme values of the component design variables such as design speed, perception-reaction time (a high percentile), and friction coefficient (a low percentile). A reliability method is presented, based on AASHTO, that does not rely on extreme values but instead considers the moments (mean and variance) of the probability distribution of each random variable. The method also accounts for correlations among the component random variables. In Cases I and II of AASHTO, the variations of the sight distance along both legs of the intersection are considered for both design and evaluation. For evaluation (involving an exiting obstruction), these variations are combined into a single variable that determines whether the corresponding sight line is obstructed. In Case III, only the sight distance leg along the major road has variations. The proposed method is straightforward and involves simple, closed-form mathematics for calculating sight distance and associated reliability. Sensitivity of ISD to various design variables is examined. ISD reliability-based values for various cases are presented from data reported in the literature, and results are compared with current AASHTO design values.