The ultimate resolution of electron optical systems far exceeds that of conventional electron resists. Electron beams less than 0.5 nm in diameter can be produced but the smallest features that can be written with these beams in standard resists are about 10 nm–20 nm in size and the minimum spacing for closely spaced lines is about 40 nm. The reason for this loss of resolution is not clear. It may be due to the range of the Coulomb interaction between the beam electrons and the resist molecules, due to the straggling of secondary electrons away from the primary beam, due to the high molecular weight of the resist or due to the mechanism of the development process. In the first two cases resolution would improve if the photon energy required to effect exposure was increased. With PMMA, the most studied resist, the energy required for exposure is about 5 eV. We have been working with an alternative nanolithography process using SiO2layers that has about three times better resolution than PMMA. We assume that the resolution is better because the photon energy required to effect exposure is higher than it is with PMMA. THe process relies on the electron beam induced enhancement of the etch rate of SiO2in buffered hydrofluoric acid and requires a much heavier exposure dose than PMMA. The process was discovered in the 1960s but its ultra‐high resolution was only identified last year in our Cambridge laboratory. SiO2is used in mamy semiconductor devices so the method should be readily applicable to device fabrication. This paper reviews the resolution limits of the various nanolithography techniques including ‘hole‐drilling’ and contamination resist and compares them with the SiO2process.