Despite its importance, is evapotranspiration poorly studied in páramo ecosystems. This study assesses the performance of 30 models, including 21 empirical models, (radiation-, temperature-, combination- and mass transfer-based), 8 artificial neural network models (ANNs), and 1 multivariate adaptive regression spline (MARS) model for the estimation of daily reference evapotranspiration (ETo) in comparison to the standard Penman-Monteith equation (FAO 56 P-M). An additional objective was to define for the study region the best alternative to the standard method. Available and limited data of two weather stations, respectively Toreadora (2013-2016 period) and Zhurucay (2014 period), both located in the páramo ecosystem of the Azuay province, in Southern Ecuador, were used. Simple statistical metrics (MBE, MAE and RMSE) were applied to evaluate the performance of the models. A random forests analysis was carried out to define the relevance of the weather variables in the evapotranspiration process. The random forest results were used for assembling the ANNs using different combinations of weather variables. This approach permitted to define the ANN with the smallest number of inputs that best estimate ETo. The MARS model enabled to derive an empirical equation, called REMPE, which uses solar radiation and minimum relative humidity as variable inputs. From the group of empirical equations, the combination-based equations have the best performance followed by the radiation-, temperature- and mass transfer-based equations. A calibration method was applied to improve the performance of the tested models. Results showed that the improved ANNs are the most accurate for estimating daily ETo, while the REMPE equation, despite been developed under local conditions, presents low performance. The annual ETo was calculated for all the models and compared against the annual value computed with the FAO 56 P-M equation. Overall, results permit to select the best model as a function of the availability of weather data in super-humid environments such as páramo ecosystems.