The reaction between a labelled monoclonal antibody (MoAb) and its specific target is characterised by three parameters: the association constant (Ka) of the labelled MoAb, the number (N) of effective binding sites on the specific target, and the immunoreactive fraction (F) of the labelled MoAb preparation. Immunological binding parameters are usually estimated graphically, by fitting the experimental data to linear equations derived from the first order law of mass action (FLMA) at equilibrium. However, only two parameters can be estimated simultaneously in a two-dimensional plot. Consequently, graphical estimation of Ka, F and N must be performed stepwise, using at least two different plots. The three parameters are interdependent, and therefore a stepwise estimation procedure might give suboptimal results. In order to investigate whether this is a problem of practical significance in the evaluation of labelled MoAbs, a computerised iterative nonlinear least squares (INLSQ) method was applied to estimate the three parameters simultaneously. The binding parameters in reactions between different 125I-labelled MoAbs and different types of targets were significantly changed when a graphical procedure was replaced by the computerised INLSQ method, and the goodness of fit to FLMA was improved. Hence, the nonlinear least squares method is the preferred procedure. Values were affected when only a subset of the data was included in the estimation procedure, indicating some heterogeneity even in these presumably homogeneous MoAb reactions. The radiolabelling procedure was presumed to be the main reason for this heterogeneity.