Reference evapotranspiration (ET
0) is an essential component in hydrological and ecological processes. The Penman–Monteith (PM) model of Food and Agriculture Organization of the United Nations (FAO) model requires a number of meteorological parameters; it is urgent to develop high-precision and computationally efficient ET
0 models with fewer parameter inputs. This study proposed the genetic algorithm (GA) to optimize extreme learning machine (ELM), and evaluated the performances of ELM, GA-ELM, and empirical models for estimating daily ET
0 in Southwest China. Daily meteorological data including maximum temperature (
Tmax), minimum temperature (
Tmin), wind speed (
u2), relative humidity (RH), net radiation (
Rn), and global solar radiation (
Rs) during 1992–2016 from meteorological stations were used for model training and testing. The results from the FAO-56 Penman–Monteith formula were used as a control group. The results showed that GA-ELM models (with R
2 ranging 0.71–0.99, RMSE ranging 0.036–0.77 mm·d
−1) outperformed the standalone ELM models (with R
2 ranging 0.716–0.99, RMSE ranging 0.08–0.77 mm·d
−1) during training and testing, both of which were superior to empirical models (with R
2 ranging 0.36–0.91, RMSE ranging 0.69–2.64 mm·d
−1). ET
0 prediction accuracy varies with different input combination models. The machine learning models using
Tmax,
Tmin,
u2, RH, and
Rn/
Rs (GA-ELM5/GA-ELM4 and ELM5/ELM4) obtained the best ET
0 estimates, with R
2 ranging 0.98–0.99, RMSE ranging 0.03–0.21 mm·d
−1, followed by models with
Tmax,
Tmin, and
Rn/
Rs (GA-ELM3/GA-ELM2 and ELM3/ELM2) as inputs. The machine learning models involved with
Rn outperformed those with
Rs when the quantity of input parameters was the same. Overall, GA-ELM5 (
Tmax,
Tmin,
u2, RH and
Rn as inputs) outperformed the other models during training and testing, and was thus recommended for daily ET
0 estimation. With the estimation accuracy, computational costs, and availability of input parameters accounted, GA-ELM2 (
Tmax,
Tmin, and
Rs as inputs) was determined to be the most effective model for estimating daily ET
0 with limited meteorological data in Southwest China.
Full article