Recent studies have proposed the use of person-based frameworks for the optimization of traffic signal timing to minimize the total passenger delay experienced by passenger cars and buses at signalized intersections. The efficiency and applicability of existing efforts, however, have been limited by an assumption of fixed cycle lengths and deterministic bus arrival times. An existing algorithm for person-based optimization of signal timing for isolated intersections was extended to accommodate flexible cycle lengths and uncertain bus arrival times. To accommodate flexible cycle lengths, the mathematical program was redefined to minimize total passenger delay within a fixed planning horizon that allowed cycle lengths to vary within a feasible range. Two strategies were proposed to accommodate uncertain bus arrival times: (a) a robust optimization approach that conservatively minimized delays experienced in a worst-case scenario and (b) a blended strategy that combined deterministic optimization and rule-based green extensions. The proposed strategies were tested with numerical simulations of an intersection in State College, Pennsylvania. Results revealed that the flexible cycle length algorithm could significantly reduce bus passenger delay and total passenger delay, with negligible increases in car passenger delay. These results were robust to changes in both bus and car flows. For bus arrival times, the robust optimization strategy seemed to be more effective at low levels of uncertainty and the blended strategy at higher levels of uncertainty. The anticipated benefits decreased with increases in the intersection flow ratio because of the lower flexibility of signal timing at the intersection.