Dulcitol, known in a single crystalline form (hereafter called I), in fact exhibits complex thermal behaviour. A second polymorph II appears when the glassy material is allowed to crystallize at room temperature. This form always coexists with I and cannot be well characterized. The mixture I + II transforms very slowly to pure I when annealed at room temperature for a week. When a melt of dulcitol is cooled very slowly vitrification does not occur and a new pure crystalline form III is obtained. Analysis of X-ray powder data leads to a possible monoclinic cell with a = 8.82 angstrom, b = 11.41 angstrom, c = 8.61 angstrom, beta = 110.55-degrees, Dx = 1.49, Z = 4. These parameters are very close to those of I but the slightly smaller values for a and b could indicate a shrinking of some bonds in the molecular packing. Each of the two forms II (mixed with I) and pure III are thermodynamically unstable. The ''normal'' form I is always obtained by evaporation from methanol/water solutions.