[en] Identification of the dominant factors involved in carbide fuel fission gas swelling and release behaviour has been attempted by comparison between experimental data and the results predicted by means of a physical model. The model assumes that fission gas release occurs entirely through gas atom migration in the matrix solid and that fission gas bubbles, intra and intergranular, grow as the net result of gas atom precipitation into the bubbles and gas atom re-solution from the bubbles. Further, it is assumed that local gas atom redistribution process in the immediate neighbourhood of a bubble is so rapid that the bubble size always corresponds to the equilibrium size that maintains exact balance between the rate of resolution and that of precipitation. Computation runs performed with the model using carefully chosen combination of physical parameters have successfully reproduced the spread of experimental gas release and swelling data. Comparisons between the predicted results and the experimental data readily identify the grain size, and not the temperature, as the dominant factor affecting fission gas behaviour. The effect of other fuel design parameters such as fission rate density, hydrostatic pressure, etc. is generally shown to be minor. Further study, however, indicates that the external fuel dimensional changes resulting from fuel cracking very often overshadow that from fission gas swelling alone. It is concluded that efforts to control carbide fuel swelling should be directed towards the control of fuel microstructure rather than the control of fuel porosity as has been generally practiced so far. (author)