Atmospheric water vapour residence time (WVRT) is an essential indicator of how atmospheric dynamics and thermodynamics mediate hydrological cycle responses to climate change. WVRT is also important in estimating moisture sources and sinks, linking evaporation and precipitation across spatial scales. In this Review, we outline how WVRT is shaped by the interaction between evaporation and precipitation, and, thus, reflects anthropogenic changes in the hydrological cycle. Estimates of WVRT differ owing to contrasting definitions, but these differences can be reconciled by framing WVRT as a probability density function with a mean of 8–10 days and a median of 4–5 days. WVRT varies spatially and temporally in response to regional, seasonal and synoptic-scale differences in evaporation, precipitation, long-range moisture transport and atmospheric mixing. Theory predicts, and observations confirm, that in most (but not all) regions, anthropogenic warming is increasing atmospheric humidity faster than it is speeding up rates of evaporation and precipitation. Warming is, thus, projected to increase global WVRT by 3–6% K⁻¹, lengthening the distance travelled between evaporation sources and precipitation sinks. Future efforts should focus on data integration, joint measurement initiatives and intercomparisons, and dynamic simulations to provide a formal resolution of WVRT from both Lagrangian and Eulerian perspectives.