Poultry feed was contaminated artificially with either Salmonella kedougou or S. livingstone using a two-stage mixing process. Intestinal infection became established in a small proportion of birds when feed containing between 0.1 and 0.3 salmonellas/g was given continuously for 2 or 3 weeks from the day of their purchase as 'day-olds' while nearly all birds became infected when the feed contained 100-300 salmonellas/g. Between these limits a dose response was demonstrated in that the proportion of birds becoming colonized with salmonellas increased as the numbers of salmonellas added to feed was increased. Nalidixic acid-resistant strains of both serotypes were used to facilitate the recovery of organisms. The isolation rate was higher from dilutions of caecal contents inoculated directly onto brilliant green agar supplemented with nalidixic acid than it was from swabs of cloacal faeces, even when an enrichment technique was used. This observation confirms that the incidence of salmonella carriage in flocks will be under-estimated if only cloacal faeces are cultured. Of the two serotypes used S. kedougou proved the more efficient colonizer although for both serotypes variation in infection rates was demonstrated in different groups of birds given feed containing comparable numbers of salmonellas.