Prior models for prediction of permeability in MMMs such as Maxwell? Bruggman? Higuchi? Pal and other models are based on the consideration of only two distinct phases of polymer and non-organic particles in its structure. Thereby, models based on consideration of three phases such as modified Maxwell? Felske and modified Pal models were developed. These models, besides of the mentioned two phases, consider a third phase as an intermediate phase within the MMM structure. Due to existing similarity between gas permeation and heat transfer in nano-fluids, some presented models for prediction of nano-fluids effective thermal conductivity coefficient are used for prediction of MMM permeability. In this thesis, a new model based on applying different structural models for predicting effective permeability in complex particles and MMMs is proposed. It is based on their physical and structural differences. In the new established model, the effective permeability of complex particles has been calculated by co-continuous structural model that was recently proposed for prediction of effective conductivity coefficient in composite materials. In the next step, by substituting the calculated effective permeabilities of complex particles and polymer matrix into the Maxwell- Garnett model the final model for permeability of MMM was derived. By applying Bruggman differential procedure? inaccuracy of Maxwell-Garnett model for large concentration of fillers was treated. The new model considers the effects of particle shape and interfacial shell layer formed between polymer and filler particles. The prior three phase’s models are unable to calculate the interfaciale shell layer properties such as volume fraction of this phase in complex particles and its permeability. The prior models uses a curve fitting procedure for estimation the shell layer properties, while the new model make a more accurate estimation based on some assumptions for estimation of the volume fraction, and defining an experimental parameter for estimation of the permeability of various gases through interfacial shell layer without using of curve fitting procedure. The new model uses primary data including permeability of polymer and fillers, and volume fraction of fillers in MMM. For evaluation of model accuracy and its ability to predict the MMMs performance, several experimental data from literatures were used. Results show that the maximum averaged relative error between model prediction and experimental data is 10.65% for permeation of N 2 through a Ultem and Zeolite 4A MMM, while the minimum averaged relative error is 2.8% for permeation of N 2 through a matrimid and carbon molecular sieve MMM. Furthermore, in this thesis another model for prediction of effective permeability in MMMs with cylindrical shape fillers such as carbon nano tubes is proposed. For derivation of this .