We review and integrate the findings from six studies sponsored by the Department of Energy's Detection and Analysis of Naturally Fractured Reservoirs Project. The project seeks to provide cost-effective tools to expand development and production efficiency of currently economically marginal, low-permeability reservoirs. This multidisciplinary research initiative includes theory, laboratory study, geology in wellbore and outcrop, remote sensing, and particularly seismic exploration. Imaging and geophysical remote sensing provide large-scale reconnaissance of regional structural trends, fracture orientations, and basin geometry. Fundamental geological study of in-situ stress and fractures in the wellbore and outcrop is essential to designing subsequent seismic surveys and predicting permeable fracture distribution. Seismic methods can directly image faults and fractures of sufficiently large scale; otherwise subseismic fracture zones can be inferred from azimuthal anisotropy, that is, directional variations, in seismic attributes (for example, seismic waves typically travel slower across fractures than parallel to them). Shear (S-) waves can best map fracture density and orientation, even in two-dimensional (2D) lines, but are still relatively expensive and are insensitive to fracture contents (water vs. gas). Compressional (P-) waves are somewhat less sensitive to fractures, but respond also to fracture contents; furthermore P-wave methods are widely used and comparatively inexpensive. Correlations of S-wave velocity anisotropy and P-wave amplitude variation with offset (AVO) in orthogonally intersecting 2D lines strengthen the case for P-waves as a cost-effective way of detecting fractures. In three-dimensional (3D) surveys, multiazimuth processing reveals spatial variations in P-wave anisotropy, highlighting inferred fracture zones. Where a dominant structural trend exists, simple two-azimuth processing is sufficient to identify regions of high anisotropy and gas yield. In this method, raypaths are separated parallel and perpendicular (each ?45?) to the dominant fracture direction; ratios or differences of seismic attributes between these independently processed data then eliminate lithologic heterogeneity and show changes in anisotropy alone. Complex structure (faults, folds) scatters fracture perpendicular raypaths; in this case structural interpretation and analysis of some seismic attributes are optimal in the fracture-parallel direction alone. Where no dominant structural trend exists, raypaths must be divided into three or more azimuths in order to determine arbitrary directions and magnitudes of anisotropy throughout the survey. Anisotropic normal-move out, in which the number of azimuths is equal to the number of common-midpoint traces, is the most general framework for analyzing azimuthal anisotropy, but requires the highest quality data (high fold in all azimuths). For cost-effective mapping of fractures and associated gas, we recommend (1) perform adequate field and remote reconnaissance; (2) acquire 3D P wave surveys with offsets equal to or greater than target depth in all azimuths; (3) process in as many azimuths as allowed by cost; (4) calibrate with limited S-wave data (surface seismic reflection or downhole vertical seismic profiles).