Systems research has led to the conception and development of various steady-flow, chemically-reactive, engine cycles for stationary power generation and propulsion. However, the question that remains unanswered is: What is the maximum-efficiency steady-flow chemically-reactive engine architecture permitted by physics? On the one hand the search for higher-efficiency cycles continues, often involving newer processes and devices (fuel cells, carbon separation, etc.); on the other hand the design parameters for existing cycles are continually optimized in response to improvements in device engineering. In this paper we establish that any variation in engine architecture-parametric change or process-sequence change-contributes to an efficiency increase via one of only two possible ways to minimize total irreversibility. These two principles help us unify our understanding from a large number of parametric analyses and cycle-optimization studies for any steady-flow chemically-reactive engine, and set a framework to systematically identify maximum-efficiency engine architectures. (C) 2014 Elsevier Ltd. All rights reserved.