Background: Heavy metal pollution has significant impacts on health and ecosystems; remediation technologies can reduce the cost to solve these problems. Heavy metals present a severe problem in the environment, mainly for their tendency to persist, bioaccumulate and biomagnification in the trophic chain. Removing these toxic compounds from wastewater remains a challenging task. Aim: Heavy metal removal capacity was analyzed using adsorbent pellets made with natural bentonite, kaolin, and zeolite. This study describes the equilibrium adsorption and kinetics of metal removal by using linear and nonlinear regression analysis. Adsorption mechanisms were also analyzed. Methods: The goodness of fit of the adsorption equilibrium data was tested with the four linearized forms of the Langmuir equation, as well as the Freundlich, Temkin, and Dubinin-Radushkevich models. To choose the best-fit model with greater reliability, five error functions were used: R2, X2, SSE, ABS, and ARE. For adsorption kinetics the Pseudo First Order, Pseudo Second Order and Elovich models were studied with linear and nonlinear regression analysis. Results and Discussion: Type I linearization of the Langmuir isotherm showed the best fit for the three metals, with maximum adsorption capacities for lead, copper, and cadmium of 7.27, 1.45 and 0.28 mg/L, respectively. The results show that Pseudo Second Order with linear regression best fitted for lead and copper data and Pseudo First Order model with linear regression for cadmium. Conclusions: Nonlinear regression was found better to fit adsorption equilibrium models and linear regression to fit kinetics models. The main mechanisms responsible for adsorption in the system are thought to be ion exchange between functional groups and cations and surface charge attraction related to Van der Waals forces.