A cytotoxicity assay with Chinese hamster ovary cells (CHO) capable of detecting 750 pg of pertussis toxin was assessed for use as a rapid test for the diagnosis of pertussis and compared with direct immunofluorescence (DFA). With pure bacterial cultures and simulated clinical specimens, the CHO assay detected as few as two colonies of Bordetella pertussis; no cytotoxicity occurred with other respiratory tract microorganisms. Next, nasopharyngeal aspirate secretions and nasopharyngeal cultures harvested after 72 h of incubation from 57 culture-positive and 201 culture-negative patients were examined. The CHO assay with nasopharyngeal secretions was positive in 25 (45%) of 55 culture-positive cases; DFA was positive in 15 (26%) of 57 cases (P = 0.05). The CHO assay with 72-h culture washes was positive in 42 (75%) of 57 culture-positive cases (P less than 0.001 compared with DFA). The CHO assay was more specific than DFA; all five CHO-positive, culture-negative cases were confirmed as true positives by serologic or toxin neutralization assays. In contrast, only 4 (36%) of 11 DFA-positive, culture-negative cases were confirmed as pertussis by serologic methods (P = 0.03). Combining the CHO assay with culture significantly decreased the delay in laboratory diagnosis of pertussis (3.30 versus 4.54 days; P = 0.01). The CHO assay is a sensitive and specific assay for the rapid diagnosis of pertussis.