We prove that the problem of deciding whether a 2- or 3-dimensional simplicial complex embeds into ℝ3 is NP-hard. This stands in contrast with the lower dimensional cases which can be solved in linear time, and a variety of computational problems in ℝ3 like unknot or 3-sphere recognition which are in NP INTERSECTION co-NP (assuming the generalized Riemann hypothesis).
Our reduction encodes a satisfiability instance into the embeddability problem of a 3-manifold with boundary tori, and relies extensively on techniques from low-dimensional topology, most importantly Dehn fillings on link complements.