Alphabetic working memory (WM) tests, such as the Wechsler Adult Intelligence Scale–III and IV Letter Number Sequencing, are not appropriate for nonalphabetic cultures. This study examined the psychometric properties of the Taiwan Odd–Even Number Sequencing Test (TOENST) and identified representative norms. The TOENST and other mental screening tasks were administered to 300 randomly selected healthy participants, 32 purposive sampling patients with schizophrenia, and 32 quota sampling controls. To investigate reliability and validity, a subset of the 300 healthy participants was randomly selected to receive a second TOENST (n = 30) or conventional WM tests (n = 42). The split-half reliability of the TOENST ranged from 0.69 to 0.95, and its test–retest reliability was 0.75. Criterion validity was demonstrated by significant correlations with conventional WM measures (all p < .05, except semantic verbal fluency), and construct validity was demonstrated by significant correlations with aging (main effect, F_10,259 = 10.99, p < .001). Normative data were established, and performance was significantly associated with age and education. TOENST scores of patients with schizophrenia were significantly lower and correlated with frontal lobe tests, but not demographical or clinical characteristics. The TOENST has adequate psychometric properties and clinical utility and is as a viable alternative WM task for nonalphabetic cultures.