Fluvial ecosystems are vital for biodiversity and human welfare but face increasing threats from flow intermittency caused by climate change and other human activities. To better understand drivers of flow intermittency, we analyzed long-term and spatially explicit river drying data from the Rio Grande, a regulated river in the North American desert southwest that was historically perennial but is now persistently intermittent. We examined the spatial structure and influences of precipitation, temperature, in-channel infrastructure, and river discharge on flow intermittency using multivariate autoregressive state space (MARSS) models and 12 years of daily data. Our findings indicate that river diversion rates at dams and irrigation return flows significantly structure the spatial occurrence of flow intermittency, but factors (possibly geologic) at distances less than or equal to 7 kilometers (km) are more influential as predictors of drying. Controlling influences of temperature and precipitation were not detected at the reach level (about 154 km) but were significant at each of the subreach scales (n equals 3) investigated. At all subreach scales, the effect of temperature exceeds precipitation by 2.5 times and is the strongest predictor of drying. Overall, process variance decreased by 98% between our reach- and all subreach models, suggesting that scale-sensitive models have great potential to accurately inform environmental flow management strategies aimed at mitigating negative effects of climate change and water extraction.