Gait phase detection is important for evaluating the recovering of gait ability in patients with paralysis as well as for determination of stimulation in FES walking. Forceplate in three dimensional system can give the most objective and accurate information about gait phase detection, however it is very expensive and limited in restricted area of laboratory. In this study, three different motion sensors(accelerometer, gyrosensor and tilt sensor) were used for detection of gait events(heel strike, HS; toe off TO) and they were compared with each other to find out the most applicable sensor for gait phase detection. Four hemiplegic patients(group A: cadence 40 steps/min, group B : cadence 70 steps/min) were volunteered for this study and motion sensors were attached on the shank and heel of subjects. Gait phase detection accuracy of each sensor was assessed using the result of FVA(Foot Velocity Algorithem. Gait phase detection using motion sensors were valid because they all have the reliability more than 95% compared with FVA(objective standard). HS and TO were determined by using FVA and processed motion sensor signal and detection accuracy of motion sensors were assessed by calculation of time interval of HS and TO between FVA and motion sensors. Vertical acceleration from accelerometer could detect the HS most accurately in hemiplegic patients group A. Gyrosensor could detect the HS and TO most accurately in both hemiplegic patients group A and B. In this study, we could find the most applicable motion sensors to detect te gait phases in patient with hemiplegia. It is respected that this objective and quantitative methos could help to decide the motion sensor for gait phase detection in hemiplegic patients.