Propidium monoazide (PMA) penetrates non-viable cells with compromised membranes. PMA has been proposed to improve the specificity of Xpert MTB/RIF (Xpert) for the detection of viable Mycobacterium tuberculosis. This study assessed the effect of PMA on Xpert cycle thresholds (C_T) of M. tuberculosis made non-viable under antibiotic pressure. In vitro, we measured the difference between C_T with and without PMA (ΔC_T) in liquid cultures treated with one of six anti-tuberculosis drugs (isoniazid, rifampin, pyrazinamide, ethambutol, streptomycin, moxifloxacin) and found significant ΔC_T only with isoniazid and ethambutol for pan-susceptible M. tuberculosis and only with ethambutol for extensively drug-resistant M. tuberculosis. In the clinic we assessed ΔC_T in sputum samples collected from patients with pulmonary tuberculosis before and at regular intervals over 12 weeks after initiation of treatment. Before treatment start, estimated C_T were 19.3 (95% CI: 17.1–21.4) and 19.8 (95% CI: 17.6–22.1) without and with PMA, respectively. Under treatment C_T increased by 2.54 per √√day (95% CI: 1.38–3.69) without PMA and an additional 0.55 per √√day (95% CI: 0.37–0.74; p < 0.0001) with PMA. We conclude that PMA increases the specificity of Xpert for viable M. tuberculosis but the effect is small and dependent on the antibiotics used.