OBJECTIVE To determine the sensitivity of differential avian and human delayed-type hypersensitivity skin testing in the diagnosis of non-tuberculous mycobacterial lymphadenitis.

RESULTS One hundred and twenty four patients had non-tuberculous mycobacteria isolated from lymph nodes over this period, 59 of whom had differential skin testing performed. The sensitivity of a response of ⩾ 10 mm to the avian precipitin was 58 of 59. No patient had both a negative human and avian Mantoux. The sensitivity of the human Mantoux alone for diagnosing non-tuberculous mycobacterial infection was 81% for a response of ⩾ 5 mm and 66% for ⩾ 10 mm. Ten patients had a 0 human response. Fifty five of the 59 patients had an avian response at least 2 mm greater than the human response.

CONCLUSION The avian Mantoux is a very sensitive method of diagnosing non-tuberculous mycobacterial infection in children. The human Mantoux is not sensitive enough to be used alone as a surrogate to diagnose non-tuberculous mycobacterial infection.