Researchers report the number of cycles corresponding to the lower strain rate in a permanent deformation test, known as flow number (FN), as a good asphalt mix stability indicator; however, its variability has limited its implementation. Current procedures for calculating FN rely on data smoothing techniques, which can yield very different values for the same specimen if different moving average periods are used. A new methodology for determining FN is proposed by modeling the permanent strain versus load cycles curve as a continuous function. Data from cold-mix replicate samples were analyzed and compared, and a substantial reduction in FN variability was observed using the proposed methodology. Use of FN to select the data to determine the power model parameters was also evaluated. Results indicate that establishing FN as the upper cut-off point of the data used to fit the power-model produces a slight reduction in parameter estimates variability. Adopting the proposed model could be useful for permanent deformation characterization and as a unifying criterion for selecting the data used for that purpose.