The movement of a part of human body, e.g. a limb, is monitored. The movement includes N iterated rotations in a plane. A sensing device is fixed to the limb and provides data indicative of the limb movement. These data are processed to generate a rotation signal. Then, portions of the rotation signal corresponding to each movement iteration are identified. To this purpose, a plurality of null-velocity instants is identified, wherein the angular velocity of the sensing device is null. Then, each null-velocity instant is classified as a candidate start/end time or a candidate peak time. Then, a start time, end time and peak time for each iteration are determined as a combination of two candidate start/end times and a candidate peak time that fulfils certain conditions on the values of the rotation signal in correspondence of candidate start/end times and candidate peak times.