MacConkey, eosine-methylene blue, deoxycholate-citrate, salmonella-shigella, and xylose-lysine-deoxycholate agars were compared for their ability to support the growth and to facilitate the recovery of enteroinvasive Escherichia coli strains from artificially contaminated as well as from clinical faecal samples. When grown as pure cultures, the 78 enteroinvasive Escherichia coli strains, as a group, exhibited the same growth characteristics as did Shigella isolates ( n=59), i.e. both organisms grew more weakly than did Salmonella strains ( n=22) on the various selective plates but 4- to 10-fold better than normal Escherichia coli isolates ( n=53). Xylose-lysine-deoxycholate and deoxycholate-citrate plates were more effective in recovering enteroinvasive Escherichia coli from faecal samples than was salmonella-shigella agar. Likewise, xylose-lysine-deoxycholate agar, similar to the differentiating MacConkey and eosine-methylene blue agars, was less inhibitory for defined "sensitive" strains than were the selective media tested. Preincubating clinical faecal samples in selenite F or in gram-negative broth did not influence the recovery of enteroinvasive Escherichia coli significantly. These data show that the use of xylose-lysine-deoxycholate, in combination with MacConkey or eosine-methylene blue agar, provides the best chance for recovery of enteroinvasive Escherichia coli when randomly selecting colonies from faecal cultures for subsequent molecular or immunological identification assays.