In a multihop wireless network, the traffic at each node and the channel over each link may fluctuate with time. Thus, traditional optimal resource allocation needs to be computed for each moment with instantaneous information of channel states over all links and the traffic rates at all nodes, leading to huge communication overhead and computation cost. To solve this challenge, in this correspondence, we propose to use robust resource allocation, in which the only needed information is the mean and variance of the wireless channels and the traffic rates. In the formulated problem, there are probabilistic constraints, which are difficult to handle. Effective methods are provided that can transform the probabilistic constraints to convex constraints. As the resource allocation does not need instantaneous channel state information or instantaneous traffic rate information, it is robust to channel and traffic variations, with very little communication and computation overhead.