In an equilibrium chemical reaction mixture, the number of molecules present obeys a Poisson distribution. We report that, surprisingly, the same is true of a large class of nonequilibrium reaction networks. In particular, we show that certain topological features imply a Poisson distribution, whatever the reaction rates. Such driven systems also obey an analog of the fluctuation-dissipation theorem. Our results shed light on the fundamental question of when equilibrium concepts might apply to nonequilibrium systems and may have applications to models of noise in biochemical networks.