Maintaining erect human posture depends on graviceptive information. This can come from at least of three origins: vestibular, visual and somaesthetic. We hypothesize here that subject's use proprioception rather than visual or vestibular cues for their control of upright body posture and this even when subjects stand on a tilting body support surface. In order to find experimental evidence for this hypothesis, we exclude in our experiments visual cues (eyes close) and by keeping frequency and amplitude of the tilt stimulus so low that it would be below the detection threshold for vestibular semi-circular canal stimuli. The orientations of body segments were analysed during various phases of the perturbation cycle. Segmental stabilisations were defined in terms of both the global anchoring index calculated during the whole perturbation cycle and an appropriate sequential anchoring index calculated during various phases in the perturbation cycle. We show that subjects tend to align their bodies with the space vertical and do so better for their heads than for their upper bodies and lower bodies. A further finding is that stabilisation is related to the tilt stimulus in the form that it is minimal at the turning points of the tilt, where peak tilt velocity is minimal with the sinusoidal stimulus used. These finding suggest first that proprioceptive cues are predominant in the control of body orientation in quasi-static condition and second that the head and trunk stabilisation strategies used as the basis of postural control depend on the properties of the moving support.