Mycobacterium abscessus has recently emerged as the cause of an increasing number of human infections worldwide. Unfortunately, it is highly resistant to existing drugs, and new specific agents to combat M. abscessus have not yet been found. The discovery of antibiotics that are effective not only against replicating but also against dormant and often recalcitrant cells is a daunting challenge. In this study, we developed a model of non-replicating M. abscessus, which represents a valuable screening tool for antibacterial agents. Thus, we demonstrated that, under a deficiency of potassium ions in the growth media and prolonged incubation, M. abscessus entered a ‘non-culturable’ state with a significant loss of colony-forming ability, but it retained viability, as confirmed using the most-probable-number (MPN) assay. The ‘non-culturable’ mycobacteria possessed decelerated cellular metabolism and noticeable differences in cell morphology from actively growing mycobacteria. ‘Non-culturable’ cells were used in a comprehensive screening of the efficacy of antibiotics, along with actively growing cells. Both CFU and MPN tests confirmed the prominent bactericidal effect of moxifloxacin on actively growing and ‘non-culturable’ M. abscessus, as proven by less than 0.01% of cells surviving after antibiotic treatment and prolonged storage. Bedaquiline exhibited a comparable bactericidal effect only on metabolically inactive non-culturable cells aged for 44 days. There were reductions ranging from 1000 to 10,000-fold in CFU and MPN, but it was not so efficient with respect to active cells, resulting in a bacteriostatic effect. The demonstrated specificity of bedaquiline in relation to inert non-replicating M. abscessus offers a new and unexpected result. Based on the findings of this research, moxifloxacin and bedaquiline can be regarded as potential treatments for infections caused by M. abscessus. In addition, a key outcome is the proposal to include the combination of viability assays for comprehensive testing of drug candidates. Relying on CFU-based assays alone resulted in overestimates of antibacterial efficacy, as demonstrated in our experiments.