Cardiopulmonary capacity should be evaluated accurately to determine exercise intolerance and training intensity of stroke survivors before an exercise rehabilitation programme is prescribed. However, no cardiopulmonary exercise test (CPET) is suitable because of the stroke victims' physical impairment. The aim of this study was to develop and validate a new rowing-ramp protocol as a CPET for stroke survivors. Eleven stroke patients (6 male; 5 female; age, 45 + 16.01 years, performed two incremental exercise tests on a Concept II rowing ergometer to determine the peak oxygen consumption (VO 2 peak). Test-retest reliability for VO 2 peak, measured 1-week apart, resulted in an intra-class correlation of 0.97 and 0.95, respectively. A linear regression equation was developed to predict the VO 2 peak from final stage stroke power. Validity and reliability of the prediction equation were established. The regression equation for predicted VO 2 peak was VO 2 peak=11.429±+ 0.232 (Final Stage Stroke Power) + 12.63 (F=25.326, p<0.01; R=0.859, R 2 =0.738). Limits of agreement between predicted and measured VO 2 peak were acceptable, with a mean bias of 0.37 ml/kg/min. The validity coefficient (R) was 0.83 (p<0.01) and 0.81 (p<0.01) in both trials. Test-Retest reliability coefficient for predicted VO 2 peak 0.95 (p<0.01). The positive relationship between Final Stage Stroke Power and VO 2 peak suggests that the Rowing-Ramp protocol could be used to measure VO 2 peak of stroke survivors. Additional studies are needed to cross-validate the regression equation using larger sample size, different type and severity of stroke.