The only known in planta substitution of starch is phosphorylation. Whereas the function of starch phosphorylation is poorly understood, phosphorylated starch possesses improved functionality in vitro. Molecular models of native crystalline starch are currently being developed and the starch phosphorylating enzyme has recently been discovered. Accordingly, it is desirable to obtain a more exact description of the molecular structures of phosphorylated starch. We have determined the crystal structure of methyl alpha-D-glucopyranoside 6-O-phosphate as its potassium salt which is thought to be the starch phosphate counterion in vivo. From this structure and previously known glucophosphate structures we describe the possible 6-O-phosphate geometries and through modeling extrapolate the results to the double helical structure of the crystalline part of amylopectin. The geometries of the existing crystal structures of 6-O-phosphate groups were found to belong to two main adiabatic valleys. One of these conformations could be fitted into the double helical amylopectin part without perturbing the double helical amylopectin structure and without creating steric problems for the hexagonal chain-chain packing.