Colorado River water is vital to the development of the arid southwestern United States. Water allocation policies are based on historical records of river flow covering ~50 years. We are evaluating the utility of the limited historical record by reconstructing the flow of the river during the last 1000 years. We use oxygen isotope cycles in living and fossil bivalve mollusks to calculate the amount of fresh water once delivered to the Gulf of California. In the absence of any river input the d18O cycles of the living bivalve Chione range from +0.75 to -2.2 per mil PDB. Although fossil specimens have similar maximum values, minima are typically more negative, sometimes as low as -7 per mil. Because temperature cycles in shallow Gulf waters have changed little during the last 1000 years, the difference can be attributed to differences in the d18O of seawater at the Colorado's mouth. This difference is largest at times of maximum river flow during the May to July pulse of snow melt.An oxygen isotope mixing relationship between Gulf seawater (+0.8 per mil SMOW) and river water at peak seasonal flow (-15 per mil) can be used to calculate the proportion of river water and salinity of the water in which fossil Chione lived. An unusual release of Colorado River water to the Gulf of California occurred in 1997-1998; comparison of shell d18O grown during this release with that before and after will allow us to relate river water proportions to river discharge. We calculate Colorado River paleo-flow at times other than seasonal peak flow by comparing d18O cycles between fossil and living shells. The area between the two d18O cycles is a measure of the amount of Colorado River water present. A mixing relationship anchored to a seasonally variable d18O value for Colorado River water is used to calculate total annual flow. This method permits us to compare the frequency distribution of paleo-flows with one derived from historical observations.