Abstract. The hypothesis that ray parenchyma cells are actively involved in the refilling of embolized xylem of Pinus sylvestris L. was tested by killing the ray parenchyma and comparing rehydration of killed stems with that of control material. Killing of ray parenchyma was achieved using hot water or sodium azide. In most experiments, the available water for refilling was at negative water potential. Experiments were done on three kinds of plant material: small branch segments, potted seedlings and small potted trees. In all experiments, there was no indication that the azide-killed xylem was slower to refill than the control material and it was concluded that the parenchyma has no role in the refilling process, which therefore must be purely physical or physicochemical. Stems treated with hot water did not refill; we suggest that this may be caused by high temperatures decreasing the water permeability of the tracheid wall. The refilling of small branch segments may be explained by surface tension forces (capillary action), which inside the tracheid lumen may lower the water potential down to -9.7kPa; this may be enough to draw in water from the available water that in experiment one was at -2kPa. In the case of seedlings or saplings, capillary action cannot explain refilling, because the xylem water potentials were always lower than those estimated from tracheid radii. Condensation of water during diurnal cycles of warming and cooling is also unlikely to contribute to refilling significantly. To account for refilling in these cases, it is supposed that the tracheid wall may be chemically active and able to lower the water potential below the value expected by capillarity.