Molecular discreteness is apparent in small-volume chemical systems, such as biological cells, leading to stochastic kinetics. Here we present a theoretical framework to understand the e ects of discreteness on the steady state of a monostable chemical reaction network. We consider indepen- dent realizations of the same chemical system in compartments of di erent volumes. Rate equations ignore molecular discreteness and predict the same average steady-state concentrations in all com- partments. However, our theory predicts that the average steady state of the system varies with volume: if a species is more abundant than another for large volumes then the reverse occurs for volumes below a critical value, leading to a concentration inversion e ect. The addition of extrinsic noise increases the size of the critical volume. We theoretically predict the critical volumes and verify by exact stochastic simulations that rate equations are qualitatively incorrect in sub-critical volumes.