Allied health personnel and nonanesthesiologist physicians often undergo training in tracheal intubation but then may actually use the skill relatively infrequently. This study assessed retention of skills one year after initial training and identified specific areas of knowledge critical to successful performance of intubation. Eleven respiratory therapists on the staff of a 253-bed hospital, each of whom had been trained one year previously in airway management, were evaluated. Prior to returning to the operating room for skills assessment and recertification, each respiratory therapist took a 21-question written exam. Therapists then went to the operating room and a trained observer (anesthesiologist) monitored the intubations performed to see whether critical steps were followed, while a second observer monitored a checklist of skills performed. The attending anesthesiologist recertified the therapist only when all steps were correctly performed and the intubation was successful. There was a poor correlation (r = -0.25, p > 0.1) between the number of intubations performed by the therapists for emergencies in the previous year and the number of intubations needed to be recertified. There was a negative correlation (r = -0.8, p < 0.05) between the score on the written test and the number of intubations required for recertification-a higher score meant fewer intubations were needed to achieve recertification. First-pass success occurred significantly more frequently if all skills tested were performed correctly (50/75 first-pass successes had all skills performed correctly vs 10/28 for failed first-pass, p < 0.01). The most common errors were levering the blade on the upper teeth (12/91) and tube not inserted from the right side of the mouth (28/104). When the blade was levered, 8 of 10 intubations failed. When the tube was not inserted from the right side of the face, 6 of 12 failed. The useful findings of this study are: (1) occasional performance of intubation did not ensure skill maintenance; (2) cognitive and procedural abilities correlated, suggesting benefits to study as well as to practical training; and (3) two specific mistakes were associated with a high incidence of failure.