The determination of an ODF, C-coefficients, property tensors and portions of texture components from EBSD orientation measurements is afflicted with statistical errors introduced by incomplete sampling of the grains. Since the measurements are highly spatially correlated and stochastically dependent, classical sampling theory does not apply. A general statistical :method for error estimation in the presence of stochastically dependent observations has been developed and applied to the most important quantities of texture analysis. The method is based on the assumption of a finite range of dependence between different measurements and on the estimation of the covariance in the observed set of orientation. The methods allows the computation of standard measurement errors and confidence limits for the mentioned texture quantities. It can be used for an objective decision whether two textures are statistically equal or not, based on the comparison of estimated ODFs. Further we can decide statistically whether the ODF obeys certain types of symmetry (e.g. whether it is a girdle textures or whether it is symmetric about the shear plane observed in the field).