Introduction The presented simulator of human rhythmic movements is based on the original model of central pattern generators (CPG), which generate periodic signals for controlling the muscles of the joint. Current models of parkinsonian movements do not simulate lower structures that obtain signals from decision-making structures in brain and are involved in controlling rhythmic muscle activity. Namely, rhythmic activities are influenced by Parkinson disease the most. The aim of this work is to simulate human lower limbs to study the effects of Parkinson's disease on walking. Global simulator would consist of three consequent levels-computational models: basal ganglia, network of CPGs, and musculoskeletal lower limbs. The presented work covers second and third levels regarding single hip joint. Simulator architecture The CPG model consists of four layers for four types of neurons. Output signals of a pair of motoneurons are fed to the inputs of the Hill models of two opposite leg muscles. Muscles pull thighbone and musculoskeletal model calculates feedbacks, which are represented by three types of sensory neurons. The presented work is illustrated by implementation of neuro-musculoskeletal system of right human thigh, controlled by two muscles of the hip joint, Iliopsoas and Glutei. Three types of sensory neurons react to speed of contraction, force, and length of muscles. They affect motoneurons of both own and opposite half-centers. Modelling results CPG is able to rhythmically control muscles without upper control from brain. The latter only influences modes of CPG. Phase diagram of the joint showed stable swings after 4 periods of transition from stillness to one swing per 2 seconds. The change in the oscillation frequency of the CPG showed a short transition period before stabilization of the hip swings. The phase diagram of the hip joint showed stabilization of periodic motion on a single trajectory for each speed value. Conclusions Simulation shows symmetrical and smooth swings, with amplitude limited to values of natural gait. Used version of control is presented for the first time. The further work is directed on creation of a stable human gait with CPG models for each pair of muscles for both legs. Keywords: central pattern generator, musculoskeletal model, Parkinson’s disease