We present a study of the dissociative chemisorption of NO, O2, and N2 over close-packed, stepped, kinked, and open (fcc {111}, {211}, {311}, {532}, {100}, and {110}) transition metal facets using density functional theory (DFT). The offset of the Brønsted–Evans–Polanyi (BEP) relations suggest that the {111} surface is the least reactive, and that the {110} surface is the most reactive. This observation is verified by establishing volcano-relations based on mean-field microkinetic models for each facet, showing that the maximum rate over {110} is 4 orders of magnitude larger than the maximum {111} rate. The ordering of the maximum activity over the facets is: {110} > {100} ~ {532} > {311} ~ {211} > {111}, which is in general agreement with the offset in the BEP relations. We show that the top-point location and shape of the volcano relations are approximately independent of facet. This observation lends credibility to the approach of analyzing trends in catalytic reactivity over a single low-index facet, and assuming the experimentally observed activity trends are qualitatively well-described by such a single-facet analysis. Our study suggests that a key element for generally obtaining quantitative agreement between theory and experiments is for the simulations to address in detail the propensities of the various types of active sites. Finally, we show that the ordering of NO decomposition rates among metals and facets is essentially unaltered when using BEP- and scaling relations in the microkinetics instead of explicit DFT calculations for each elementary reaction step, and that using a “universal” BEP relation introduces no significant additional qualitative error in trend prediction.