The water balance method is used, in Brazil, to estimate the out ow from the regularization reservoirs, based on hourly data. The the rating curve method, normally adopted in conventional stream ow gauge stations of the hydrometeorological network national, the values are estimated from the average of two daily readings of the water level elevation and requires the use of extrapolation techniques to the estimation of values outside the validation limits. The objective of study was to analyze the variation of the reservoir's hourly out ows Queimado Hydroelectric Power Plant (HPP Queimado) and to propose a method to evaluate whether the estimate of the daily out ows, obtained by the water balance method, are similar to the ow values obtained at a stream ow gauge station of the ANA network. The logistic regression model was used because it is a method that adopts binary categorical dependent variables to identify the event of interest. The results showed that the average values of stream ow obtained between the observations made from the average of 2 or 24 readings daily were close, so that the estimated data for two daily readings show good representation regarding the variability of the ows in the region. The water balance and rating curve methods showed similar statistical performance for most of the stream ow data considered. The logistic regression was able to identify atypical data in the series of analyzed ows, especially in the rainy season, which are mainly associated with the methods for extrapolation of the rating curve.