Water sorption has been studied gravimetrically for polychloroprene rubber samples, first at a fixed hygrometric ratio (98% HR) and several temperatures (25, 40, 60 and 80 °C) for samples of 1.8 and 3.8 mm thickness (Constant Temperature and Hygrometry, CTH experiments), then at fixed temperature (40 °C) and several hygrometric ratios ranging from 0 to 95% HR on samples of 0.1 mm thickness (DVS experiments). CTH experiments reveal an abnormal sorption behavior: after an apparently fickian transient period, the water absorption continues at almost constant rate, no equilibrium is observed after more than 2500 h, whatever the temperature. DVS experiments reveal a very low Henry's solubility but the formation of clusters at water activities higher than 40%. The water diffusivity is almost independent of activity below 50% HR and decreases rapidly when activity increases above 50%. Contrary to CTH experiments, equilibrium is reached in DVS and the difference is not simply linked to the well-known effect of sample thickness on diffusion rate. The results allow hypotheses such as hydrolysis or osmotic cracking to explain the abnormal sorption phenomenon to be rejected. It is suggested that clusters could be polymer–water complexes having a linear/branched structure able to grow without phase separation that could explain the reversibility of sorption–desorption cycles. The difference of behavior between thin 0.1 mm and thicker 1.8 or 3.8 mm samples could be due to an effect of swelling stresses.