A comparative study of yield to intermediate products obtained from the distributed-feed membrane reactor (MR) and the conventional cooled-tube plug-flow reactor (PFR) is presented. The nonisothermal, nonadiabatic one-dimensional pseudohomogeneous reactor model was used. The analysis is presented in terms of the usual dimensionless groups (Damkohler number (Da), heat of reaction parameter, Stanton number (St), etc.) and delineates regions of parameter space where the membrane reactor gives improved yield with no higher maximum temperature. Optimal conditions for the PFR usually lie on or close to the boundaries between either pseudo-adiabatic operation (PAO) or hot-spot operation (HSO), and runaway operation. Results show that when compared under optimal operating conditions and at the same catalyst volume, reactant feed rate, temperature and pressure, the membrane reactor performance is better than that of the PFR only over restricted regions of parameter space. In general, distributed feed is preferred at higher Da, and under conditions of high heat release or lower cooling rates. The distributed-feed reactor performance can be enhanced by a mixed-feed strategy, or by slightly increased inlet temperature.