"In X-ray structure determination, the R-value reports how well a model agrees with the experimental data. In small molecule crystallography, R-values of 3% are routinely reached. However, for biological macromolecules, R-values are normally around 20%, even when data extend to atomic resolution. This is clear evidence that our current models of macromolecular crystal structures are severely and generally lacking. But what is the difference between our models and reality? And how can we improve our models with this information? Are we extracting enough information from these data? Macromolecular crystals contain large regions of disordered solvent –a soup of ions, buffers, PEG and water– which contribute to the measured reflection intensities. They also affect the ordered macromolecules, which adopt slightly different conformations depending on local environment resulting in an ensemble of similar structures rather than one. Because we cannot determine these features as accurately as we would like, in particular low resolution reflections and phases are difficult to model. Phasing(structure solution) and refinement of macromolecular X-ray structures suffer from this lack of understanding. We aim to improve structure determination and model quality with model-free phases and electron density derived from MAD experiments. New and more sophisticated scale functions and solvent models will be tested; phases obtained by different approaches will be analysed for strengths and weaknesses. Improved phase estimates will result in better models, and hence more reliable biological conclusions; all macromolecular structures will benefit. In particular the structure determination of membrane proteins and large macromolecular complexes will be enhanced, where often only mid-low resolution data are available and the poor phase estimates currently obtainable result in noisy electron density maps. The gained information may also influence other methods such as NMR, SAXS, FEL and EM."