A fundamental thesis of seismic stratigraphy is that seismic reflections follow impedance contrasts that coincide with stratal surfaces, which are surfaces where depositional processes occur at a fixed moment in geologic time. This stratal-surface concept is used herein to image a narrow (width--300 ft), thin, fluvial channel system that is embedded within a seismic reflection peak. The peak reflects from a large (about 2- x 2-mi) area where nonchannel facies dominate its waveshape. The targeted channel facies are confined to an interval that vertically spans less than 30 ft. According to principles of seismic stratigraphy, four conformable seismic stratal surfaces that pass through the interior of this channel sequence were constructed across a 2- x 2-mi (approximately) area of a 3-D seismic-data volume. The channel images portrayed on these seismic horizons, which were spaced at vertical increments of 2 ms, illustrate the principle that seismic attributes viewed on seismic stratal surfaces provide valuable images of facies distributions within thin-bed sequences and help seismic interpreters segregate channel facies from nonchannel facies. These stratal-surface interpretations of the fluvial system were then integrated to create a thin (8-ms-kck), stratal-bounded seismic-data window that spans the short geologic time period during which the targeted fluvial depositional system was active. Seismic-attribute calculations within this stratalbounded analysis window improve parts of the channel image by integrating the facies information from all stratal surfaces into a unified seismic attribute that vertically spans the total depositional sequence. A comparison is made between channel images on seismic stratal surfaces that are conformable to two different reference surfaces, one reference surface being positioned below the targeted fluvial system and the second reference surface being above the h-bed channels. This comparison supports the premise that seismic interpreters should extrapolate stratal surfaces both upward and downward across a thin-bed target to optimize the image of that target.