Microbial contamination of complementary foods is a major cause of childhood diarrhoea. In a community-based study in Ghana, we evaluated whether fermentation of maize porridge or storage of porridge in vacuum flasks reduces coliform contamination. The complementary food examined, Weanimix, consisted of, on a dry weight basis, toasted maize (75%), peanuts (10%), and soybeans (15%) milled into flour. The fermented food was Weanimix made with drum-dried, previously fermented maize. Fifty women with infants 6-18 months of age participated. A repeated measures cross-over design was used. Each mother participated in all four treatments (1 week per treatment, in random order): (a) non-fermented Weanimix (W), (b) fermented Weanimix (F), (c) non-fermented Weanimix stored in a vacuum flask (WV), and (d) fermented Weanimix stored in a vacuum flask (FV). Each week, mothers were supplied with the appropriate food, and asked to prepare a porridge each morning by boiling the dry mix in water. Samples were collected in the evenings and immediately plated onto 3 M Petrifilm. Contamination was defined as > or = 100 colony-forming units of coliforms per ml. Contamination rates (95% confidence intervals) were W = 48% (38-58%), F = 25% (17-34%), WV = 42% (32-52%), and FV = 13% (6-20%). All pairwise comparisons were significant (P < 0.05) except for W vs WV. Within the WV treatment, contamination rates were 85% when the sample temperature fell below 50 degrees C (N = 41) vs 12% when it remained > 50 degrees C (N = 59) (P < 0.001). These results indicate that contamination is reduced by fermentation, and further reduced by vacuum flask storage of fermented foods. For non-fermented foods, vacuum flask-storage was protective only when the temperature was maintained at > 50 degrees C; at < or = 50 degrees C vacuum flask storage increased the risk of contamination.