Attention is called to some little-discussed aspects of crystal structure prediction. It is pointed out that the structures of metal-organic frameworks can often be correctly predicted. This allows designed synthesis of materials with targeted pore size and functionality. For more-conventional inorganic materials, predicting the lowest energy state can be made difficult by two problems. First, the difference in energy between competing structures can be very small. Second, the unit cells of materials with simple compositions may be very large. Examples are given for both cases.