A descriptive mathematical model was chosen to fit the antigen-antibody association kinetics of a new homogeneous immunometric assay for prolactin, involving time-resolved fluorescence detection (TRACE technology, Time Resolved Amplified Cryptate Emission). We paid special attention to the methodology and criteria applied, to yield a convenient and statistically valid model, designed to allow potential exploitation of kinetic information in the data processing of the assay. We compared specific parameterizations of an hyperbolic model, the Gompertz, and the monomolecular models on the basis of morphological considerations, a statistical analysis of fit, and an assessment of the parameters estimation quality, over a wide range of antigen concentrations. The monomolecular model gave the best fit, and the most precise and stable estimation of its parameters. The study of parameter properties confirmed this choice.