Dual right-turn lanes are increasingly used as a design alternative at urban intersections, primarily to accommodate high right-turn demand. For dual right-turn lanes, an accurate estimate of right-turn-on-red (RTOR) capacity can contribute to better decisions about whether RTOR should be allowed or prohibited and may lead to refined delay estimation and improved signal timing. A gap-acceptance model was formulated for predicting lane-specific RTOR capacities at dual right-turn lanes. The proposed model can represent the unequal effects of conflicting traffic streams from different cross-street lanes on RTOR capacities of dual right-turn lanes. Existing probabilistic methods were adapted to adjust RTOR capacity for shared through and right-turn lanes. Microsimulation models were developed, calibrated based on field data, and used as benchmarks to validate the proposed model. Numerical experiments indicated that the proposed model exhibits a significantly improved ability to predict RTOR capacities for dual right-turn lanes compared with the classical Harders model.