Four industrial beryllium grades were tested via nanoindentation. An extremely high variation of hardness was observed in all samples. Analysis of the nanoindentation data in combination with SEM/EBSD measurements demonstrated that the crystallographic orientation of the indented grain was the major source of the wide variation in hardness, which was 2.5 times higher when the indentation direction was close to the  c-axis of beryllium compared to indentation along the [11-20] or [1-100] directions. Crystal plasticity finite-element (CPFEM) simulations indicated how this hardness anisotropy arises from the anisotropy in the plastic deformation. Experiments and simulations also demonstrated that localised plastic deformation of the surface around the indent (pile-up or sink-in) was highly crystallographically dependent: during indentation into “soft” orientations, pile-up dominated; while sink-in behaviour was dominant during indentation into “hard” orientation. This implies that the hardness values calculated from indenter displacement and indenter profile using the standard Oliver-Pharr approach, without considering pile-up/sink-in effects, will be incorrect. Several contact area correction methods were applied and are compared. In contrast the indentation modulus was not found to have any strong crystallographic dependence. CPFEM analysis indicates that this is due to the complex 3-dimensional nature of the elastic interaction between the indenter and the sample, and also since, for the chosen indentation depth, the elastic interaction volume is much larger than the materials’ grain size.