A novel model is developed for optimizing the phased development of a pre-designed rail transit line. The investment plan and extension phases of the line are optimized over continuous time and under budget constraints to maximize net present value (NPV) over an analysis period. This model determines the maximal allowable train headway while considering demand elasticity. This model is first formulated for a one-directional extension problem, and then modified for its two-directional version. A genetic algorithm with customized operators is developed for optimizing the sequence and grouping of station completions. For each problem version the model is demonstrated with a numerical case and its corresponding optimized solution. The sensitivity of results to several important input parameters is analyzed. Results show that potential demand and in-vehicle time value greatly influence the optimized NPV, while unit construction cost and potential demand are most influential on the optimized extension plan.