This study aims to compare multivariate calibration methods developed from data obtained by square wave anodic stripping voltammetry using a hanging mercury drop electrode for simultaneous determination of metals in cachaça, the following metals were studied: copper, zinc and cadmium. Multivariate calibration, partial least squares (PLS) and artificial neural network (ANN) methods were used in previous studies using other electrodes for this determination. In this new study, besides ANN and PLS, a hybrid model that combines PLS and NN, namely PLS-Neural was used. Also, samples of industrial cachaças were incorporated into the study in addition to artisanal samples. The quality of the methods was evaluated in terms of coefficient of determination (R2) and root mean square error of prediction (RMSEP). F test was used for comparing methods at confidence level of 95%. Based on these studies, it was found that although all methods show good results, the method employing neural networks stands out in the determination of copper in samples of cachaça. All methods proved to be fast and relatively low-cost, and they can be used for such analyses.