In many models of inflation, the period of accelerated expansion ends with preheating, a highly non-thermal phase of evolution during which the inflaton pumps energy into a specific set of momentum modes of field(s) to which it is coupled. This necessarily induces large, transient density inhomogeneities which can source a significant spectrum of gravitational waves. In this paper, we consider the generic properties of gravitational waves produced during preheating, perform detailed calculations of the spectrum for several specific inflationary models, and identify problems that require further study. In particular, we argue that if these gravitational waves exist they will necessarily fall within the frequency range that is feasible for direct detection experiments—from laboratory through to solar system scales. We extract the gravitational wave spectrum from numerical simulations of preheating after λ⁴ and m²² inflation, and find that they lead to a gravitational wave amplitude of around Ω_gwh²~10⁻¹⁰. This is considerably higher than the amplitude of the primordial gravitational waves produced during inflation. However, the typical wavelength of these gravitational waves is considerably shorter than LIGO scales, although in extreme cases they may be visible at scales accessible to the proposed BBO mission. We survey possible experimental approaches to detecting any gravitational wave background generated during preheating.