The objectives of this study were to evaluate the role of trauma to the skin in development of Escherichia coli cellulitis and to compare the abilities of three cellulitis isolates (O78, O115, O21,83), one airsacculitis isolate (untypable) and one fecal isolate (O86) of E. coli to induce cellulitis in broiler chickens. Forty-eight 4-week-old commercial broiler chickens were housed in groups of six in eight battery cages. For five groups, the skin on the left side of the abdominal region of chickens was traumatized by scratching with a 22-gauge needle, then contaminated with a swab dipped in a broth culture of one of the five E. coli isolates. For chickens in the remaining three groups, an avian cellulitis culture (O115, O21,83) or sterile broth was applied to intact skin. The experiment was duplicated. All birds were euthanatized 10-13 days postinoculation. No lesion developed in chickens in which the skin had not been traumatized. Among the traumatized birds, cellulitis isolates induced characteristic lesions of cellulitis in 86% of the birds, whereas airsacculitis and fecal isolates induced lesions in 42% and 8% of birds, respectively. Severe or moderate gross pathologic changes were found in 86% and microscopic pathologic changes were found in 88% of birds inoculated with cellulitis isolates; the corresponding percentages for the airsacculitis isolate were 25% and 17%. This study demonstrated that trauma to the skin is necessary for initiating disease and that strains of E. coli of serotypes epidemiologically associated with cellulitis are highly virulent in experimental infection.