We study the approximation of functions that map a Euclidean domain Ω ⊂ R d into an n-dimensional Riemannian manifold (M,g) minimizing an elliptic, semilinear energy in a function set H ⊂ W 1,2 (Ω,M). The approximation is given by a restriction of the energy minimization problem to a family of conforming finite-dimensional approximations S h ⊂ H. We provide a set of conditions on S h such that we can prove a priori W 1,2 -and L 2approximation error estimates comparable to standard Euclidean finite elements. This is done in an intrinsic framework, independently of embeddings of the manifold or the choice of coordinates. A special construction of approximations -geodesic finite elements-is shown to fulfill the conditions, and in the process extended to maps into the tangential bundle. Key words and phrases. Geometric finite elements, L 2 -error bounds, vector field interpolation Energy minimizing maps into and between Riemannian manifolds arise in many contexts, both theoretical and applied. Existence results for harmonic maps have consequences for curvature and topology [23]. Isoperimetric regions (minimizing the area functional), for example, with large volume center in manifolds asymptotic to Schwarzschild have been explored in the context of general relativity and the ADM mass [16]. In general, the influence of an ambient geometry has been of growing interest in the context of geometric flows like mean curvature flow, Ricci flow and Willmore flow. More applied examples are the modelling of oriented materials in Cosserat theory [34], liquid crystals [1], and micromagnetics [29]. Manifold valued harmonic map heat-flow has also been introduced as a regularization in image processing [47].The research interest extends beyond energy minimization in ambient Riemannian manifolds. In ambient spacetimes, spacelike hypersurfaces with vanishing mean curvature are maximizers of the area functional and play a role in general relativity as initial data for solving the Einstein equations. When considering limits of smooth manifolds or problems in optimal transport, the manifold structure of the ambient space needs to be replaced by the general framework of metric spaces. While these prospects are certainly interesting, we will limit our attention to smooth Riemannian target manifolds in the hope that later adaptations can provide a more general theory.