As part of the mechanistic-empirical pavement design procedure, the flow number test is used to rank the bituminous mixtures for rutting. In the flow number test, a sample is subjected to 0.1 s of loading followed by 0.9 s rest period and a haversine load waveform is used. This test protocol has serious limitation and these are due to the loading conditions applied and the post-processing procedure followed for flow number determination. In this investigation, three changes were made in the existing flow number protocol. The first one was in the load waveform. A trapezoidal waveform was used instead of the haversine load waveform. The second change was related to loading conditions. Different combinations of confinement and deviatoric loading were applied and the final choice was made based on the resulting three stage creep curve. The third change was related to post-processing method and a Francken model to capture the complete three stages of deformation was used. A BC mix with mid-gradation following MoRTH (Specifications for roads and bridge works. Indian Roads Congress, New Delhi 2013) guidelines was used with VG30 and CRMB60 binder. PReSBOX shear compactor was used for making cylindrical samples of 94 mm diameter and 150 mm height. The samples were cast with a target air voids of 6 %. All the testing was carried out at 50 and 60°C. Different confinement conditions ranging from 50 to 200 kPa were used with the deviatoric loading ranging from 200 to 600 kPa. The influence of modified binder was clearly seen in the development of three stage curve. The collected data was post-processed using Francken model and the flow number was determined. This flow number was bench marked with the existing flow number protocol and it was seen that there were gross discrepancies. Keywords Flow number Á Three stage creep curve Á Confinement pressure Á Viscoelasticity Á Rutting Á Francken model Transp. in Dev. Econ. (2016) 2:10