Bacterial cultures were made on tube-feeding formulas provided to 35 unselected adult patients on termination of the infusion. Bacteria were counted and identified using routine procedures. Formulas were classified as nonmanipulated, manipulated, or locally prepared. Medical records were reviewed to determine if diarrhea was present during the period that included 2 days on either side of the sampling day. A significant association was observed between the extent of bacterial contamination and the presence of diarrhea (p = 0.027). Locally prepared and manipulated formulas contained a significantly greater number of organisms when compared to nonmanipulated formulas (x2 = 17.81, p less than 0.001). Counts for two baseline and four termination cultures on locally prepared formulas exceeded the acceptable limit for coliform organisms in pasteurized milk according to public health standards. Eight additional cultures on commercial feeding formulas which were presumably sterile initially, exceeded these standards at the end of the administration period. Thus 12 of 35 formulas (34%) would not meet public health standards applicable to pasteurized milk. Use of sterile nonmanipulated formulas in a closed administration set is recommended, along with routine, periodic bacteriologic surveillance of enteral feeding programs.