We evaluated therapeutic plasma exchange (TPE) efficiency in treatment of a single relapse in steroidrefractory patients with neuromyelitis optica spectrum disorders (NMOSD) in a multi-ethnic resource-limited setting. This was a historical cohort study on the clinical outcomes post-TPE in a multiethnic cohort of 53 steroid-refractory NMOSD patients at a single Malaysian tertiary center. Primary outcomes, assessed both pre-and post-TPE, were Medical Research Council scale of muscle power, Modified Rankin Scale, Expanded Disability Status Scale (EDSS), and visual acuity. Secondary outcomes were ambulatory status and target neurological deficit (TND)-based TPE response. Significant improvements in Medical Research Council, Modified Rankin Scale, EDSS, and visual acuity (P < 0.001) were observed at 1-month post-TPE with further improvement of EDSS at 6 months (EDSSΔ6) post-TPE (P < 0.001). At 6 months post-TPE, 87% of patients has successful TND-based TPE response and 69.8% were ambulating without support. Patients with anti-aquaporin 4 seronegativity (P = 0.004), myelitis and brainstem features at first relapse (P = 0.004), longer cord lesions (P = 0.030), higher pre-TPE EDSS of ≥8 (P = 0.018) and delayed TPE initiation of >14 days (P = 0.047) were significantly associated with improved EDSSΔ6. TND-based TPE response was significant in absence of cord atrophy (P = 0.030). TPE is an effective treatment for steroidrefractory acute relapses of NMOSD in a multiethnic Malaysian population despite its resource-limited setting. The predictive factors of EDSSΔ6 improvement were antiaquaporin 4 seronegativity, longer cord lesions, and higher pre-TPE EDSS. Absence of cord atrophy was predictive of better TND-based TPE response. Unexpectedly, our study showed that delayed TPE initiation of more than 14 days and up to 60 days may also be beneficial.