A comment on the article entitled "Molecular dynamic simulation of the vanadium pentoxide gel host" [J. Linde and J.O. Thomas, Solid State Ionics 85 (1996) 1] is presented concerning the model structure of the V2O5 layer of vanadium pentoxide gels. The double-layer structure used in the simulation was derived from the crystalline V2O5. This model structure significantly differs from the double-sheet type determined by the X-ray Rietveld method. The simulation should be recalculated based on the double-sheet type V2O5 layer.