We recently reported molecular dynamics simulations of the two native forms of crystalline cellulose [A. P. Heiner, J. Sugiyama and 0. Teleman, Carbohydr. Res., in press]. From these a number of molecular properties have been calculated and show that local structure differs in triclinic and monoclinic cellulose, but also between the so-called odd and even subphases of monoclinic cellulose. Although all glucose rings remain close to the 4C1 chair conformation, pucker analysis shows that the odd subphase tends towards the 2E envelope. Literature data were systematised into a relation between certain torsion angles and solid state l3C NMR chemical shifts. Torsion angles were found to be important for the C-6 chemical shift but not for the C-1 and C-4 carbn atoms.