In recent years micromechanical devices have been developed that can strongly couple to light, by integrating them within optical cavities. A main goal has been to cool the devices optomechanically, freezing out all thermal vibrations, so that the object's motion eventually becomes limited by quantum mechanical fluctuations. This would make it possible to study a new range of quantum behaviour of mechanical objects. Thompson et al. report an improved design of such a system, involving a movable membrane sandwiched between two rigid high-quality mirrors. In previous designs one of the mirrors had to double-up as a microresonator. The new device achieves substantial cooling, from room temperature to 6.8 mK. It should eventually be possible to reach the quantum-limited ground state with this system. A report on an improved design of an optomechanical system in which a movable membrane is placed between two rigid high-quality mirrors, as opposed to previous designs where one of the mirrors has a double function as the microresonator; it's claimed that it is feasible to reach the quantum-limited ground state with this new design. Macroscopic mechanical objects and electromagnetic degrees of freedom can couple to each other through radiation pressure. Optomechanical systems in which this coupling is sufficiently strong are predicted to show quantum effects and are a topic of considerable interest. Devices in this regime would offer new types of control over the quantum state of both light and matter1,2,3,4, and would provide a new arena in which to explore the boundary between quantum and classical physics5,6,7. Experiments so far have achieved sufficient optomechanical coupling to laser-cool mechanical devices8,9,10,11,12, but have not yet reached the quantum regime. The outstanding technical challenge in this field is integrating sensitive micromechanical elements (which must be small, light and flexible) into high-finesse cavities (which are typically rigid and massive) without compromising the mechanical or optical properties of either. A second, and more fundamental, challenge is to read out the mechanical element’s energy eigenstate. Displacement measurements (no matter how sensitive) cannot determine an oscillator’s energy eigenstate13, and measurements coupling to quantities other than displacement14,15,16 have been difficult to realize in practice. Here we present an optomechanical system that has the potential to resolve both of these challenges. We demonstrate a cavity which is detuned by the motion of a 50-nm-thick dielectric membrane placed between two macroscopic, rigid, high-finesse mirrors. This approach segregates optical and mechanical functionality to physically distinct structures and avoids compromising either. It also allows for direct measurement of the square of the membrane’s displacement, and thus in principle the membrane’s energy eigenstate. We estimate that it should be practical to use this scheme to observe quantum jumps of a mechanical system, an important goal in the field of quantum measurement.