The ability of lactic acid bacteria (LAB) and Saccharomyces cerevisiae to remove aflatoxin B1 (AFB1) from liquid medium was tested. The experimental results indicated that (i) AFB1 binding to microorganisms was a rapid process (no more than 1 min); (ii) this binding involved the formation of a reversible complex between the toxin and microorganism surface, without chemical modification of the toxin; (iii) the amount of AFB1 removed was both toxin- and bacteria concentration-dependent; and (iv) quantitatively similar results were obtained with viable and nonviable (heat-treated) bacteria. According to these details, a physical adsorption model is proposed for the binding of AFB1 to LAB and S. cerevisiae, considering that the binding (adsorption) and release (desorption) of AFB1 to and from the site on the surface of the microorganism took place (AFB1 + S <--> S - AFB1). The model permits the estimation of two parameters: the number of binding sites per microorganism (M) and the reaction equilibrium constant (K(eq)) involved, both of which are useful for estimating the adsorption efficiency (M x K(eq)) of a particular microorganism. Application of the model to experimental data suggests that different microorganisms have similar K(eq) values and that the differences in toxin removal efficiency are mainly due to differences in M values. The most important application of the proposed model is the capacity to select the most efficient microorganism to remove AFB1. Furthermore, it allows us to know if a modification of the adsorption efficiency obtained by physical, chemical, or genetic treatments on the microorganism is a consequence of changes in M, K(eq), or both.