Blood pressure is one of the most common techniques used to diagnose heart disease. The blood-pressure waveform gives vital information about the blood flow through the body. In this project an auscultatory method is used to obtain non-invasive human blood-pressure signals. An algorithm based on wavelet transform is introduced to detect every single Korotkoff pulse of blood pressure.The aim is to automatically determine the systolic and diastolic points of bloodpressure measurement for (1) a subject at rest, (2) a subject walking with his/her arms still and (3) a subject walking with his/her arms swinging. With the multiscale feature of wavelet transform, pulses of blood pressure can be distinguished from noise and artifacts. The approach of the algorithm has been developed using Matlab engineering software. Detection techniques were developed to de-noise the signal, identify the blood-pressure pulses and finally to determine the systolic and diastolic points.