Therapy of multidrug-resistant (MDR)-TB is highly problematic; that of extensively drug-resistant (XDR)-TB even more so. Both infections result in high mortality, especially if the patient is coinfected with HIV or presents with AIDS. Selection of therapy for these infections is limited and, for most situations, it is performed 'blind'. However, there is a solution for the selection of effective therapy and this is presented herein. Ideal therapy of the patient infected with MDR-TB or XDR-TB can be determined a priori by the mycobacteriology laboratory. This would involve the isolation of the patient's macrophages, the phagocytosis of the mycobacterial isolate and the presentation of the antitubercular agent to the macrophage-bacterium complex. This system is reviewed in its entirety and its potential and feasibility are supported by hard experimental demonstrations.