In vitro experiments confirmed that a 10-mer peptide derived from human mammary-associated serum amyloid A3 (M-SAA3) protected intestinal epithelial cells from enteropathogenic Escherichia coli (EPEC) adherence. The entire 42-mer human M-SAA3 protein was even more effective, reducing EPEC binding by 72% relative to untreated cells (P<0.05), compared with 25% and 57% reductions for the human 10-mer and Lactobacillus GG, respectively. However, none of the M-SAA3 peptides reduced Salmonella invasion in vitro (P>0.05). Each of the M-SAA3 10-mer peptides and the 42-mer was then administered orally to mice at 500 mug day(-1) for 4 days before deliberate infection with either Citrobacter rodentium (mouse model of EPEC) or Salmonella Typhimurium. None of the peptides protected against Salmonella infection and the 42-mer may even increase infection, as there was a trend towards increased Salmonella counts in the liver and small intestine in 42-mer-treated mice compared with those in sodium acetate-treated control mice. Citrobacter counts were reduced in the caecum of mice administered the 42-mer relative to a scrambled 10-mer (P<0.05), but not compared with the sodium acetate control and no reductions were observed in the faeces or colon. Overall, although promising anti-infective activity was demonstrated in vitro, neither the 42-mer M-SAA3 protein nor a 10-mer peptide derivative prevented enteric infection in the animal models tested.