ABSTRACT Evaluating the comparative performance of U.S. federal agencies is difficult, particularly since both tasks and missions vary so dramatically. In addition, forces beyond an agency's control (e.g., COVID, an economic downturn, etc.) can determine outcomes even when agencies are performing at a high level. In this paper, we introduce a new approach to measuring organizational performance, something conceptually distinct from, but correlated with, both organizational inputs and outcomes. This measurement approach focuses on how well the internal machinery of agencies is functioning. We analyze a vast trove of subjective and objective performance information and aggregate it using a Bayesian structural equation measurement (BSEM) model. We isolate organizational performance from inputs and outcomes through careful model specification, information from the BSEM models, and model identification through a careful evaluation of different models and diagnostics. Our analysis yields 2479 organizational performance estimates for 135 U.S. federal departments and agencies spanning 19 years between 2002 and 2024. We explore the validity of these estimates by comparing them with other measures of similar or related concepts. We conclude by discussing the implications of our measurement approach and its usefulness for evaluating organizational performance in diverse and changing contexts.