Ephemeral-fluvial sandstones form the main reservoirs in many oil and gas fields. Production histories and development difficulties suggest these deposits are more complex than is commonly supposed in the petroleum industry. There is inadequate published information to account for this observation. Relevant sedimentological literature on such deposits is qualitative and unsuitable for detailed reservoir characterization studies; therefore, integrated sedimentological, petrophysical, and fluid-flow simulation studies have been conducted on outcrop analogs in the southwestern United States. Our key objectives were to define flow units and to find areas of sensitivity inherent both in the sediments and in the analytical and statistical procedures used to describe and model such systems. We report results of integrated studies on the Lower Jurassic Kayenta Formation in southeastern Utah. The Kayenta Formation is a relatively proximal, sand-rich succession dominated by channel-fill sandstones. Sedimentological analysis identified eight main genetic units. In contrast, reservoir simulations, based on probe-permeameter data collected at outcrop, show that six of the main genetic units have uniform flow behavior. Intraformational mudclast conglomerates are the dominant permeability barriers, both because of the clasts themselves and because they act as sites of preferential cementation. Stochastic permeability models conditioned to the outcrop data produce results little different from models based on averaged values for each facies. The most sensitive feature of the models concerns the assumptions made about the mudclast conglomerates. The model results show that only by integrating analytical approaches can we understand the full character of an outcrop analog in such a way that we can use it predictively.