Magnetic nanodiscs are fabricated by depositing cobalt onto 10-30 nm diameter silicon nanopillars, which were prepatterned using gold colloids as etch masks. The magnetic anisotropy energy of individual nanodiscs is studied by measuring the angular dependence of switching field using the micro-SQUID technique. The Stoner-Wohlfarth model, describing the magnetization reversal by unifom rotation, is used to analyse the data. The switching astroids of pure Co exhibit a cubic magnetocrystalline anisotropy indicating that the Co crystallites are fcc. After controlled oxidation of the nanoparticles, the anisotropy is dominated by a defect-induced uniaxial anisotropy, which means that the anisotropy can be used as a quality gauge.Understanding and engineering magnetic anisotropies is critically important in magnetic nanostructures, since they govern both the static and dynamic physical properties of such systems. The effective anisotropy, which determines the easy and hard magnetization directions is a superposition of different contributions including magnetocrystalline, surface and shape anisotropy [1]. Some of these terms can be tailored, such as the magnetocrystalline anisotropy by choosing the material and the shape anisotropy by selecting a geometry. Other anisotropy terms are related to defects, such as misf t dislocations and local oxidation spots. These anisotropies tend to be uniaxial, while the other ones can be engineered to be of higher order, such as cubic. A direct measurement of the individual anisotropy contributions is usually not possible in nanoparticle systems. However, the entire anisotropy function is measurable via the three-dimensional critical surface of the switching f elds [2]. Since our measurement of the famous Stoner-Wohlfarth astroid [3] on a single nanoparticle, we have investigated different particles from 40 nm barium ferrite to 3 nm cobalt nanoparticles [4][5][6][7][8]. One of the intriguing questions that emerged when we tried to engineer the anisotropies was why the observed anisotropies always had a prevalent uniaxial anisotropy. This observation was contrary to expectations that small fcc Co particles, which we studied, should exhibit a pure cubic magnetocrystalline anisotropy. For applications, the key task is designing anisotropies, since for applications such as memory devices, the anisotropies determine the minimum bit size which is thermally stable [9,10]. Such design is possible if the anisotropy is determined by the choice of materials and crystallographic phase rather than by defects, which are inherently hard to control. Thus the key question is whether the observed uniaxial anisotropy stems from defects or whether it corresponds to the inherent anisotropy of the system. To unambiguously demonstrate that the anisotropies can be engineered in single domain particles, and are not defect dominated, a system has to be devised which exhibits a pure higher-order anisotropy. Even for the simplest case, which is the cubic anisotropy, so far no results on single domain pa...