Electron crystallography can be used to determine the structures of membrane proteins at near-atomic resolution in some cases. However, most electron crystallography projects remain at a resolution around 10 Å. This might be partly due to lack of flatness of many two-dimensional crystals. We have investigated this problem and suggest single particle processing of locally averaged unit cells to improve the quality and possibly the resolution of three-dimensional maps. Applying this method to the secondary transporter melibiose permease we have calculated a three-dimensional map that is clearer and easier to interpret than the map derived using purely electron-crystallographic methods.