Ferroelectric and ferromagnetic materials exhibit long-range order of atomic-scale electric or magnetic dipoles that can be switched by applying an appropriate electric or magnetic field, respectively. Both switching phenomena form the basis of non-volatile random access memory(1), but in the ferroelectric case, this involves destructive electrical reading and in the magnetic case, a high writing energy is required(2). In principle, low-power and high-density information storage that combines fast electrical writing and magnetic reading can be realized with magnetoelectric multiferroic materials(3). These materials not only simultaneously display ferroelectricity and ferromagnetism, but also enable magnetic moments to be induced by an external electric field, or electric polarization by a magnetic field(4,5). However, synthesizing bulk materials with both long-range orders at room temperature in a single crystalline structure is challenging because conventional ferroelectricity requires closed-shell d(0) or s(2) cations, whereas ferromagnetic order requires open-shell d(n) configurations with unpaired electrons(6). These opposing requirements pose considerable difficulties for atomic-scale design strategies such as magnetic ion substitution into ferroelectrics(7,8). One material that exhibits both ferroelectric and magnetic order is BiFeO3, but its cycloidal magnetic structure(9) precludes bulk magnetization and linear magnetoelectric coupling(10). A solid solution of a ferroelectric and a spin-glass perovskite combines switchable polarization(11) with glassy magnetization, although it lacks long-range magnetic order(12). Crystal engineering of a layered perovskite has recently resulted in room-temperature polar ferromagnets(13), but the electrical polarization has not been switchable. Here we combine ferroelectricity and ferromagnetism at room temperature in a bulk perovskite oxide, by constructing a percolating network of magnetic ions with strong superexchange interactions within a structural scaffold exhibiting polar lattice symmetries at a morphotropic phase boundary(14) (the compositional boundary between two polar phases with different polarization directions, exemplified by the PbZrO3-PbTiO3 system) that both enhances polarization switching and permits canting of the ordered magnetic moments. We expect this strategy to allow the generation of a range of tunable multiferroic materials.