A multivariate version of the classical univariate standard addition method is proposed for the analysis of samples generating overlapping signals in the presence of notorious matrix effects. Unlike previous versions based on multivariate calibration by partial least-squares (PLS), the proposed strategy takes advantage of a self-modeling methodology: multivariate curve resolution by alternating least-squares (MCR-ALS) enhanced with signal shape constraints based on parametric functions. In this way, there is no need for the full multivariate response of a blank solution, and in multianalyte determinations, the standard additions can be made with a solution containing all of the analytes, which constitutes a clear advance as compared to PLS approach. The proposed method has been successfully tested in the voltammetric determination of hydroquinone and catechol in solutions of increasing complexity and appears to be a promising tool in the field of electroanalysis.