Macromolecular crystallographers are united by problems that plague the molecular replacement technique (multiple copies of a molecule, high crystallographic or non-crystallographic symmetry, and low homology). J. Navaza discussed combating the problem of high non-crystallographic symmetry, by exploiting the prior information available from self-rotation and Patterson functions. In complementary work, R. Read showed that likelihood functions give a much clearer indication of the correct answer than conventional scores, particularly for poor or incomplete models. Two talks described phase information that can be gained by averaging multiple images of a structure, either through noncrystallographic symmetry or multiple crystals. K. Cowtan described combining data from a crystal at room temperature and the same crystal at cryogenic temperatures, showing how changes in cell dimensions can give useful phase information. J. Wang presented an example where 14-fold non-crystallographic symmetry was used to generate an excellent map from nearly random phases. Finally, Q. Shen gave a lucid explanation of 3-beam X-ray diffraction and the use of a reference beam geometry to collect 3-beam data. In favourable cases, the resulting profiles give clear indications of triplet phases, which can be used in conjunction with direct methods programs to give clear electron density maps.