Non-perennial streams constitute a large portion of global river systems. They are defined by periods of drying and rewetting, but these drying patterns are variable and often poorly understood. Non-perennial streams are also important ecosystems that can support diverse aquatic communities and allow organism dispersal. Many studies of non-perennial streams have described the environmental and anthropogenic drivers of flow intermittency at gauged sites across broad spatial scales, but these studies provide little information about regional variation in drivers of intermittency at ungauged sites. This study combines existing spatial data with new field observations to characterize intermittence at the reach scale. One of the goals of this project is to create a model to predict stream intermittency under different climate change scenarios. The first step in creating such a model is to determine the environmental factors that affect drying patterns. Here we present a study of intermittent streams across 5 river drainages in southeastern Oklahoma. We collected flow data from 44 streams, deploying time-lapse cameras to measure water levels in riffles and Stream Temperature, Intermittency, and Conductivity (STIC) loggers in pools to track drying. We used machine learning algorithms to classify sites based on their drying regimes and statistical models to identify environmental drivers of stream drying. Characterizing drying regimes is a necessary first step to linking variation in drying attributes to variation in aquatic community structure and other ecological responses. The results of this analysis will also inform a predictive model of drying regimes for unobserved streams. We know that environmental factors, such as land use, shape drying patterns. This project has the potential to influence management by promoting land use practices that maintain water in non-perennial streams.