The numbering-up of microchannel reactors definitely faces great challenge in uniformly distributing fluid flow in every channel, especially for multiphase systems. A model of stochastic differential equations (SDEs) is proposed based on the experimental data recorded by a long-term optical measurement to well quantify the stochastic trajectories of gas bubbles and liquid slugs in parallel microchannels interconnected with two dichotomic distributors. The expectation and variance of each subflow rate are derived explicitly from the SDEs associated with the Fokker-Planck equation and solved numerically. A bifurcation in the trajectory is found using the original model, then a modification on interactions of feedback and crosstalk is introduced, the evolutions of subflow rates calculated by the modified model match well with experimental results. The established methodology is helpful for characterizing the flow uniformity and numbering-up the microchannel reactors of multiphase system. (c) 2015 American Institute of Chemical Engineers.