Redundancy allocation is common in reliability engineering. There has been extensive researches on this topic, but the study on systems that simultaneously incorporate active, warm, and cold redundancy with Weibull distribution is less. Accordingly, the reliability-redundancy allocation problem (RRAP) on this system is focused on in this paper. The reliability of the series-parallel system with active, cold standby and warm standby redundancy is analyzed using the counting process. Based on the reliability model, the optimization model is established with the goal of system reliability maximization under the constraints concerning cost and weight. Further, a genetic algorithm is used to solve this model. The proposed method is applied to a modified benchmark problem. Results from the model computation and sensitivity analysis reveal that this study makes the RRAP benchmark problems more reliable and provides system designers with a wider flexibility.