We have coupled electrochemistry, surface science, and density functional theory (DFT) calculations to quantify structure-property relationships in well-defined oxygen evolution reaction (OER) electrocatalysts. The OER is an essential component in electrochemical CO2 reduction, H2 evolution, and NH3 synthesis reactions. Mixtures of nickel, cobalt, and iron have shown remarkable alkaline OER activity, but atomic-level understanding of how OER activity evolves with the catalyst shape, size, composition, and support are still lacking in the literature. Our approach uses ultra-high vacuum techniques to grow well-defined transition metal electrocatalysts with precisely controlled morphology, composition, and loading. Scanning tunneling microscopy (STM), X-ray photoelectron spectroscopy, and X-ray absorption spectroscopy determine the catalyst’s size, aspect ratio, thickness mass loading, and oxidation state. Electrochemical OER testing then correlates mass activity, turnover frequency, and specific current density with the catalyst structure and composition. Our results with ultrathin, two-layer thick (2L) Fe2O3 nanocatalysts showed a linear relationship between the population of edge-site Fe atoms and overall OER activity, and edge-site Fe atoms are approximately 150 times more OER active than Fe atoms on the catalyst surface. In fact, 2L-Fe2O3 nanocatalysts with a high density of Fe edge-site atoms demonstrated better OER performance than an Ir-oxide catalyst. Post-reaction STM and XPS analysis confirmed that particles retained their structure and identified hydroxylated Fe edge-site atoms along the catalyst perimeter as key OER active sites. Complementary DFT modeling confirmed these experimental observations by predicting more thermodynamically favorable OER at Fe edge-site atoms due to beneficial modification of intermediate binding. Extension of this work to Ni/Fe and Co/Fe systems is revealing additional insight into the structure-dependent activity of bimetallic OER electrocatalysts, and our approach provides atomic-level structure-property relationships that would be extremely difficult, if not impossible, to quantify with traditional approaches.