Wireless sensor networks (WSNs) consist of low-power and energy-constrained sensor nodes, and a fundamental challenge in the design of such networks is to maximize the network lifetime. In WSNs, data collected by adjacent sensor nodes usually have spatial-temporal correlations, and data aggregation technique is often used as an effective approach to remove data redundancy. Efficient usage of data aggregation technique can significantly reduce the amount of data delivery, lower the cost of overall power consumption of the network, hence increase the network lifetime. This paper studies the optimal data delivery in WSNs that takes advantage of data aggregation and nodal power control, and presents a novel routing algorithm that maximizes the network lifetime. The algorithm uses genetic algorithm (GA) to achieve an optimal selection of aggregation points, and gradient algorithm is also used to further optimize the result. The algorithm balances the power consumption of sensor nodes, and maximizes the network lifetime. Numerical results show that the proposed approach has substantially improved the network lifetime.