Recent investigation of sedimentary facies within the upper part of the Dakota Formation in central Kansas indicates that the controversial Rocktown channel sandstone, well exposed in the east-west trending Saline River valley in Russell County, is the dominate lithofacies within a transgressive, marginal-marine deltaic complex. The sandstone body can be divided into two major subfacies, grading within 50 km from a highly sinuous, distributary channel into estuarine and delta-front sandstones. In overall geometry the lower-cross-bedded sandstone subfacies (18-22 m thick and 300-600 m wide) is more or less confined to a sinuous, elongate, broad V-shaped trough, whereas the upper flat-bedded sandstone subfacies (6-9 m thick and 900-1,500 m wide) forms an elongate, tabular-shaped unit which "caps" and appears to "spill over" the edges of the trough below. Both subfacies consist of very well- to well-sorted (0.30 to 0.40 phi-units), loosely-cemented quartz arenite; however, the lower, cross-bedded subfacies consists of fine-grained (2.50; 0.177 mm), relatively high porosity sandstone, and the upper, flat-bedded subfacies of very fine-grained (3.31; 0.102 mm), relatively low porosity sandstone. Paleocurrents indicated by high-angle cross-bedding of the lower subfacies show a unidirectional pattern parallel to sandstone-body trend, whereas ripple-bedding and low-angle cross-bedding in the upper subfacies show bimodal and polymodal patterns not parallel to sandstone-body trend. Overall variance of paleocurrent measurements around the vector resultants falls well within the range of values typical of fluvial-deltaic sands. And, whereas the lower sandstone subfacies contains only leaf impressions and carbonaceous debris, the upper sandstone subfacies contains locally abundant brackish-water mollusks and abundant, well-defined trace fossils. Paleoenvironmental interpretations of the Rocktown channel sandstone are best made using sandstone-body geometry, lateral lithofacies relationships, sedimentary structures, paleocurrents, and macroinvertebrate and trace fossil paleoecology determinations. Grain-size distribution parameters are the least reliable; however, the use of cumulative log-normal probability plots according to the technique developed by Visher (1969) appears to be superior to the use of discriminating plots as discussed by Friedman (1967) and Moiola and Weiser (1968) or to the use of CM diagrams of Passega (1957, 1964).