Sinus-track cultures were compared prospectively with cultures from operative specimens in 55 patients with chronic bone infection. There was a total of 115 operative isolates; 102 of the sinus-track isolates were identical (88.7%), showing a specificity of 95.7% and a predictive value of 90.3%. A sinus-track specimen proved to be a reliable source for the isolation of all bacteria causing chronic bone infection except Staphylococcus epidermidis, provided that material from the depths of the sinus was aspirated by syringe from an active flowing sinus and inoculated immediately on culture media. Cotton-swab sinus specimens were unreliable for the isolation of mycobacteria, and could miss many pyogenic bacteria, or contain contaminants. Tuberculous bone infection should be suspected if there is no growth of any pyogenic bacteria or if there is growth of Staphyloccocus epidermidis alone on routine aerobic and anaerobic sinus cultures. Mycobacteria can often be identified from sinus-track culture from patients in whom operative culture, histopathology and clinical examination have failed to confirm the diagnosis of tuberculosis