We determined that the oxygen isotopic composition of cellulose synthesized by a submerged plant, Egeria densa Planch., is related to the isotopic composition of environmental water by a linear function, δ18O cellulose = 0.48 δ18O water + 24.1%‰. The observation of a slope of less than 1 indicates that a portion of cellulose oxygen is derived from an isotopically constant source other than water. We tested whether this source might be molecular oxygen by growing plants in the presence of high concentrations of 18O in the form of O2 bubbled into the bottom of an aquarium. Cellulose synthesized during this experiment did not have significantly different oxygen isotope ratios than that synthesized by control plants exposed to O2 of normal 18O abundance. We propose that oxygen in organic matter recycled from senescent portions of the plant is incorporated into cellulose. Our findings indicate that paleoclimatic models linking the oxygen isotope composition of environmental water to cellulose from fossil plants will have to be modified to account for contributions of oxygen from this or other sources besides water.