To strengthen the results obtained during the experimental study presented in Part I of this article, a detailed theoretical analysis of silicon deposition from silanes in a fluidized bed has been performed. In a first step, four classical models of fluidized-bed reactors have been modified, to treat silicon deposition from monosilane, considering only reactions on surfaces. A convenient agreement has been obtained between the observed and the calculated data, particularly for the Kato and Wen model. But, this approach has been completely unable to treat the case of disilane. Consequently, for the first time, to our knowledge, chemical reactions in the gaseous phase have been included in the Kato and Wen model. This original approach, applied both to the cases of silane and disilane, has improved our understanding of deposition phenomena. More precisely, concerning fines formation, this work has confirmed the explanation given in Part I : it has shown the presence of sufficient active species at the top of the bed, so that these compounds could react on the cold walls of the reactor to give fines. It has also reinforced our rejection of the mechanism of gas nucleation inside the bed proposed by several authors. Secondly, this study has allowed us to strengthen our explanation of thermal disturbances and bed agglomeration, by the fact that the increase in particles cohesiveness seems to be due specifically to deposition mechanisms from monosilane, and not from silylene. However, investigations are still in progress on these subjects.