A simulated ward study was carried out to compare the microbiological risk of assembling and running four different enteral feeding systems for 24 h. Assembly was carried out, (i) according to manufacturers' instructions with hands either covered by disposable gloves or deliberately contaminated with a test organism, or (ii) touching both the nutrient container top and pump set connector with hands deliberately contaminated with K. aerogenes. Two of the systems were ready-to-hang types (pack and bottle), the other two required feed to be decanted from either bottles or cans. When manufacturers' instructions were followed and disposable gloves worn, organisms were only detected in feeds decanted from cans and at levels < or = 20 cfu ml(-1). However, when systems were assembled following manufacturers' instructions, but with contaminated hands, no organisms were found in either of the ready-to-hang systems but average bacterial counts in samples from systems where the feed was decanted from bottles were 1.8 x 10(3) cfu ml(-1) at 24 h and 9.3 x 10(5) cfu ml(-1) for those where feed was decanted from cans. When systems were deliberately touched with contaminated hands, no organisms were detected in any feed samples from the pack system at 24 h, while bacterial counts for the other three systems ranged from 10(1) to 10(5) cfu ml(-1). The results highlight the important role played by system design in reducing both the level and incidence of bacterial contamination of enteral tube feeds and indicate that ready-to-hang feeding systems should be the preferred choice. However, if decanting of feeds cannot be avoided then strict adherence to manufacturers' instructions and the use of disposable gloves is to be advised.