The present study investigates the coding of positions reached in a two-dimensional space by populations of muscle spindle afferents. The unitary activity of 35 primary muscle spindle afferents originating from the tibialis anterior, extensor digitorum longus, extensor hallucis longus, and peroneus lateralis muscles were recorded from the common peroneal nerve by the microneurographic technique. The steady mean frequency of discharge was analyzed during 16 passively maintained positions of the tip of the foot. These positions were equally distant from and circularly arranged around the "neutral" position of the ankle. The results showed that a same position of the foot was differently coded depending on whether it was maintained for several seconds or whether it was attained after a movement. Muscle spindle activity was increased or decreased, respectively, when the previous movement lengthened or shortened the parent muscle; the magnitude of change in activity depended on the amount of lengthening or shortening in relation to movement direction. Each muscle surrounding the ankle joint was shown to encode the different spatial positions following a directional tuning curve. Data were analyzed by using the "neuronal population vector model". This model consists of calculating population vectors representing the mean contribution of each muscle population of afferents to the coding of a particular position, and by finally calculating a sum vector. The direction of the sum vector was shown to accurately describe the direction of a given maintained position compared to the initial position. We conclude that muscle spindle position coding is based on afferent information coming from the whole set of muscles crossing a given joint. A given spatial position is associated with a stable muscle afferent inflow where each muscle makes an oriented and weighted contribution to its coding.