One-hundred and ninety-nine patients with peripheral lymph node tuberculosis were randomized to treatment with either E2H9R9, Z2H9R9 or Z2H6R6 regimens (E, ethambutol; H, isoniazid; R, rifampicin; Z, pyrazinamide; numbers denote duration of therapy in months). One-hundred and thirty-three patients were diagnosed by aspiration/biopsy leaving residual nodes, 44 were diagnosed after excision of the palpable nodes and 22 were diagnosed on clinical grounds supported by a strongly positive tuberculin test. Treatment was completed as planned by 157 patients. Eight patients required aspirations after commencing chemotherapy, seven on ethambutol and one on pyrazinamide (P = 0.005). There was no statistically significant difference between the regimens in speed of resolution of nodes, or in the percentage with residual nodes at the end of treatment, or in the numbers developing fluctuation or sinuses. Follow-up to 30 months from commencement of treatment continues in order to establish relapse rates for the regimens.