Control of multi-drug-resistant tuberculosis (MDR-TB) requires extensive, supervised chemotherapy because second-line anti-TB drugs have a narrower therapeutic range than first-line drugs. This study aimed to develop population pharmacokinetic (PK) models for second-line drugs in patients with MDR-TB, evaluate the recommended dosage regimens and, if necessary, suggest new dosage regimens. A prospective, single-centre PK study was performed on second-line anti-TB drugs in patients with MDR-TB. Moxifloxacin, cycloserine, p-aminosalicylic acid (PAS), kanamycin and other second-line drugs were administered to the patients. Plasma concentrations were analysed using ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS). Population PK models were developed using non-linear mixed effect modelling (NONMEM, Version 7.30; ICON Development Solutions, Ellicott City, MD, USA). Simulations were performed using the calculated PK parameters. The respective absorption rate constant, apparent clearance and apparent volume of distribution values were as follows: 0.305/h, 9.37 L/h and 56.7 L for moxifloxacin; 0.135/h, 1.38 L/h and 10.5 L for cycloserine; 0.510/h, 30.8 L/h and 79.4 L for PAS; and 1.67/h, 3.75 L/h and 15.2 L for kanamycin. The simulations showed that the following dosage regimens were more likely to be within the recommended concentration ranges than the raw data in this study: 200 mg of moxifloxacin once daily (QD) (patient weight <50 kg) and 400 mg of moxifloxacin QD (patient weight ≥50 kg), 500-750 mg of cycloserine QD, 4.95-6.6 g of PAS twice daily and 750-1000 mg of intramuscular kanamycin QD. These findings indicate that the recommended doses should be revised to improve the clinical outcomes of MDR-TB treatment.