A prospective clinical study in three phases was performed to determine whether it was possible that enteral diet containers could become contaminated as a result of endogenous organisms ascending retrogradely from the enteral feeding tube via the giving set, and if this did occur whether the incidence could be altered by modifying enteral delivery systems. Each phase observed patients on enteral feeding over a 48-hour study period (phase I, n = 18; phase II, n = 17; phase III, n = 18). Each patient was prescribed an enteral diet of 2 L/24 h administered by continuous pump infusion from a closed 1-L sterile diet container. Four containers were used for each patient, and one giving set was used in the 48-hour period. Diet samples were taken at 12-hour intervals: two from the giving set before and after flushing with residual diet, and one from the diet container. Phases differed only in the design of the giving set: phase I had no drip chamber, phase II had a drip chamber, phase III had a drip chamber and an anti-reflux ball valve at the distal end. Both phase I and III had greater numbers of giving set samples colonized at 24, 36, and 48 hours. In phase I, 3 of 14 sterile diet containers were colonized with greater than 10(4) colony-forming organisms/mL of diet at 48 hours. Only phase II had no organisms contaminating the containers at 48 hours. We conclude that the retrograde spread of organisms from patient to sterile diet container does occur in clinical practice, and that a drip chamber in the giving set may prevent the problem.(ABSTRACT TRUNCATED AT 250 WORDS)