Abstract: Crop coefficient curves for late rice were estimated using FAO-56 single and dual crop coefficient methods with experimental data from the "863" National Key Project on water-saving agriculture. Effects of adjusted maximum crop coefficient on dual crop coefficient were evaluated and the results of single and dual crop coefficient were tested by comparing the estimated and observed evapotranspiration in 2004. The rice crop coefficients at tilling, joint-booting, ear sprouting-flowering and milking stages under water saving irrigation conditions are 1.14, 1.49, 1.43 and 1.12, respectively. Estimated values obtained by FAO-56 single crop coefficient method are 1.1, 1.39 and 0.79 for the initial, development and last stages. The reestimated dual crop coefficient improved the precision of crop coefficient and reduced the relative error of crop evapotranspiration. Tested result showed that: relative error between the total crop evapotranspiration measured in 2004 and that estimated by using the single and dual crop coefficient varied within 12.42%～16.24% and dual crop coefficient method was the most suitable one for crop coefficient estimation.