A method to perform exercise testing for patients with hemiplegia is unavailable though over half of them have cardio-pulmonary disorders. We aimed to assess the reliability and validity of using a stepper in cardiopulmonary exercise testing (CPET) in this population.14 stroke patients with hemiplegia who failed to ride the stationary bike were included. Exclusion criteria included manual muscle strength ≦1 in the lower extremity, and conventional contraindications of CPET. They underwent CPET twice by using a stepper to evaluate test–retest reliability and validity. Additionally, 10 healthy participants underwent CPET twice on the cycle ergometer and stepper respectively.In the test–retest, the ratio of two-time difference to mean was 5.0, 3, 11.3 and 12.0% on average for peak oxygen consumption, peak heart rate (HR), anaerobic threshold and minute ventilation - carbonic dioxide production slope respectively. Cronbach's alpha coefficient of peak oxygen consumption and anaerobic threshold were 0.992 and 0.919. In the stepper exercise testing of the hemiplegic participants, the ratio of peak HR to age-predicted maximal HR was 75% on average. Peak respiratory exchange ratio (mean ± standard deviation = 1.17 ± 0.08) was not different from that of healthy controls (1.21 ± 0.09). Notably, VO2 trajectory in relation to work rate is nonlinear and different in the rest-retest.This is the first research to study CPET variables in detail using stepper in patients with hemiplegia. CPET variables associated with peak are valid and reliable; nonetheless, those with sub-maximum are not. The study provides a method to do exercise testing for the patients with hemiplegia and its notice in application.