In this paper we illustrate the use of 3-D seismic attribute studies for predicting the distribution of physical properties in the subsurface. Using a data set consisting primarily of digital logs and seismic data, we show how correlations can be made between seismic attributes and physical properties (porosity), and how those relationships can be exploited to predict the distribution of the property of interest in three dimensions. The results of these studies: (1) provide quantitative, site-specific 3-D models of physical properties that are of more use for applied studies than qualitative 2-D models commonly derived from facies modeling or sequence stratigraphic analysis, (2) are generally more geologically reasonable than studies based on geostatistics alone, (3) can provide sedimentary geologists with fundamental insights into depositional and/or diagenetic controls on the distribution of properties of interest, (4) need to be rigorously evaluated by integrating other types of data and analyses, and (5) are best thought of as supplementing, rather than replacing, conventional geologic analyses. The concepts and methods we illustrate may have application in various branches of sedimentary geology. Our study area is Appleton Field in southern Alabama. At this location we predict the 3-D distribution of porosity in carbonates of the Upper Jurassic Smackover Formation using a probabilistic neural network and a combination of four attributes. Our results suggest that porosity was best developed, and preserved, in thrombolite facies of the reef front.