The development of the analytical instruments for rapid in-field evaluation of integral surface water quality parameters is an urgent analytical task. Using three different sensor devices (optical sensor for cadmium, voltammetric sensor for lead and potentiometric multisensor system), we have explored the possibility of direct quantification of water pollution index (WPI) for contaminations induced by heavy metals. We have applied linear (partial least squares, PLS) and non-linear (kernel regularized least squares, KRLS) multivariate regression tools to construct predictive models evaluating the content of individual metals and WPI in complex aqueous media simulating surface water composition. We have also explored the potential of data fusion at different levels combining the signals from all three sensor devices. The results indicate that all the instruments retain the sensitivity towards target analytes in complex aqueous samples containing humic substances and that the direct quantification of WPI in the range from 1 to 4 is possible using the employed instruments with RMSE values around 0.14. © 2023 Elsevier B.V.