Aim The Montreal Cognitive Assessment (MoCA) has been shown to be more sensitive in detecting executive dysfunction than the Mini‐Mental State Examination (MMSE). However, it is still not known whether all the MoCA executive subtests contribute to the superior sensitivity. Thus, the present study aimed to determine how much executive abnormality was detected by the MMSE and MoCA executive subtests in a population‐based cohort of Chinese post‐stroke patients. Methods The MMSE and MoCA were collected from post‐stroke patients (within 15 days to 1 month after stroke, including ischemic stroke and hemorrhagic stroke) in 14 hospitals of northern and southern China (including 10 top‐graded hospitals and 4 community hospitals) between June 2011 and September 2013. The proportions of patients with incorrect MoCA executive subtests and the proportions of patients with incorrect MMSE executive subtests were compared. Results A total of 1222 patients (703 men and 519 women, aged 62.06 ± 10.68 and 62.76 ± 9.86 years, respectively) were recruited. The MoCA detected more patients with executive dysfunction than the MMSE (OR 15.399, 95% CI 12.631–18.773; P < 0.001). The likelihood of incorrect MMSE executive tasks increased across decreasing scores of MoCA executive tasks (P < 0.001 for trend). Compared with the MMSE three‐step command test (15.5%), the MoCA trail‐making (57.8%), abstraction (48.0%) and abstraction (measurement tool; 45.7%) detected more patients with executive dysfunction (P < 0.001), whereas the MoCA digit span forwards (4.3%) and backwards (11.6%) detected fewer patients (P < 0.001 and P = 0.005, respectively). Conclusions The MoCA executive tasks are more sensitive in detecting executive dysfunction compared with the MMSE executive tasks. Geriatr Gerontol Int 2017; ••: ••–••.