Summary: In this paper we study the transport capacity of a data-gathering wireless sensor network under different communication organizations. In particular, we consider using a flat as well as a hierarchical/clustering architecture to realize many-to-one communications. The capacity of the network under this many-to-one data-gathering scenario is reduced compared to random one-to-one communication due to the unavoidable creation of a point of traffic concentration at the data collector/receiver. We introduce the overall throughput bound of \(\lambda =W/n\) per node, where \(W\) is the transmission capacity, and show under what conditions it can be achieved and under what conditions it cannot. When those conditions are not met, we constructively show how \(\lambda = \Theta (W/n)\) is achieved with high probability as the number of sensors goes to infinity. We also show how the introduction of clustering can improve the throughput. We discuss the trade-offs between achieving capacity and energy consumption, how transport capacity might be affected by considering in-network processing and the implications this study has on the design of practical protocols for large-scale data-gathering wireless sensor networks.