To identify the source and route of verotoxin-producing Escherichia coli (VTEC) infection in humans, we tried to isolate VTEC from fresh deer dung collected from free-range animals in two parks during the period from August 1997 to January 1998. The results are presented below. 1) VTEC were isolated from 21 of 200 deer dung samples (10.5%), consisting of 15 of 100 samples (15.0%) collected in park A and 6 of 100 samples (6.0%) collected in park B, suggesting that the incidence of VTEC isolation differs depending on location. 2) With respect to typing of verotoxin, the 21 isolated VTEC strains consisted of 10 strains (47.6%) as VT1 producer, 5 strains (23.8%) as VT2 producer, and 6 strains (28.6%) as double producer of both types. 3) With respect to serogroup of the isolated VTEC strains, 2 strains belonged to O128:H2.1 strain each belonged to the O8:H10, O128:H12, and O169:HUT groups. The remaining 16 strains failed to be identified as particular serotypes. Regarding local distribution of the serotype, in park A, 1 strain each belonged to the O128:H2, O8:H10, and O169:HUT groups. The remaining 12 strains did not clearly show particular serotypes. In park B, 2 strains belonged to O128:H2, and 4 strains failed to show particular serotypes. The remaining 1 strains showed autoagglutination. In conclusion, we isolated VTEC strains from deer that showed types of toxin and serogroups identical to those of human VTEC. Therefore, VTEC found in deer dung could well be a source of VTEC-infectious diseases in humans.