When metabolic activity diverges from cell number: Interpreting redox and ATP assays across cell states and culture systems

Metabolism-based assays are widely used to estimate cell viability, proliferation and treatment response because they are inexpensive, scalable and compatible with routine plate readers. Their interpretation nevertheless depends on a conditional relationship: the amount of signal produced or recovered per living cell must remain sufficiently stable across the conditions being compared. Cell-cycle arrest, changes in cell size, hypoxia, differentiation, senescence and immune activation can violate this assumption without necessarily causing cell death. Chemical interference, reagent depletion, incomplete extraction and transport limitations in three-dimensional cultures introduce additional errors. This narrative review examines the mechanistic basis and experimental consequences of divergence between metabolic activity and cell number, focusing on MTT and related tetrazolium assays, resazurin reduction and ATP-dependent luminescence. Foundational studies are integrated with methodological and experimental developments published in 2024–2026. A simple measurement model distinguishes changes in population size, average cellular response and assay recovery, explaining why normalization to an untreated control or a highly linear calibration curve does not establish validity across treatments. Particular attention is given to medium optimization, mesenchymal stromal cells, hypoxic cultures, spheroids and anticancer screening. We propose a practical framework that pairs a validated metabolic readout with direct population measurements and a separate assessment of death or division when those outcomes are claimed. Discordant measurements should be investigated as potentially informative biology rather than automatically discarded as technical failure. Metabolic assays remain valuable when their outputs are named accurately, their calibration is validated in the relevant biological state and their interpretation is matched to the experimental question.