Low-energy neutrons are essential for the analysis and characterization of materials and magnetic structures. However, both continuous (reactor-based) and pulsed (spallation-based) sources of such neutrons suffer from low fluence. Steering and lensing devices could improve this situation dramatically, so increasing spatial resolution, detectable sample volume limits and even perhaps opening the way for the construction of a neutron microscope. Neutron optics have to date exploited either Bragg diffraction(1,2), such as bent crystals, or reflection, as in mirror(3) guides or a Kumakhov lens(4,5). Refractive optics remain an attractive alternative as they would permit full use of the beam cross-section, allow a compact and linear installation and, because of similarity to conventional optics, enable the use of commercial design and simulation tools. These advantages notwithstanding, single-element refractive optics have previously been considered impractical as they are too weakly focusing, too absorptive and too dispersive. Inspired by the recent demonstration(6) of a compound refractive lens (CRL) for high-energy X-rays, we have designed, built and tested a prototype CRL for 9-20-Angstrom neutrons by using readily available optical components : our CRL has gains greater than 15 and focal lengths of 1-6 m, well matched to small-angle neutron scattering.