Effects of sampling procedures on ability to culture Pasteurella spp. from Rocky Mountain bighorn sheep (Ovis canadensis canadensis) were examined experimentally. Sample site influenced (P less than 0.0001) recovery of P. haemolytica in adult bighorn sheep. We isolated nonhemolytic P. haemolytica from 18 of 19 tonsillar swabs and 18 of 19 tonsillar biopsies from adult sheep, yet only four of 19 nasal swabs yielded isolates. Sample handling also affected (P less than 0.0001) recovery of P. haemolytica. Nonhemolytic P. haemolytica was cultured from 14 of 19 tonsillar swabs plated directly onto blood agar, but from only two of 19 swabs stored for 24 hr in modified Stuart's medium. We detected nonhemolytic P. haemolytica at least once in bronchial aspirates from four and in nasal swabs from three of six bighorn lambs. Based on direct cultures of tonsillar swabs and/or biopsies, all 26 bighorn sheep (seven lambs, 19 adults) sampled were infected with nonhemolytic P. haemolytica; only two lambs developed pneumonia during the study period. Thirty-four of 37 nonhemolytic P. haemolytica isolates tested were biotype T; three were biotype A. Serotypes 3; 4; 3, 4 and 3, 4, 10 were identified in a subsample of 17 isolates. Our data suggest tonsillar swabs or biopsies plated directly onto blood agar and incubated immediately offer the greatest probability of recovering nonhemolytic P. haemolytica from health bighorn sheep.