We have decided to write up how we think and what we do when we are engaged in structure solution from powder diffraction data - a thing we love to do.

It is often not sufficient to rely solely on diffraction data - merging the refinement against a powder diffraction pattern with other sources of information may feel like putting together pieces of a puzzle. And the crystal structure is correct when all the pieces fit.

To illustrate the process, we use several examples with increasing complexity when it comes to extracting accurate structural information form a powder diffraction pattern. We start from the crystal structure of cytosine that could be refined with no restraints and end up with a structure where we had to decide between two very similar structure models, based on a Rietveld refinement. In between we discuss the use of databases to find the expected values for geometry parameters, how to use and rely on chemical sense in deducing a crystal structure model, how to employ difference Fourier maps in structure completion and validation and how to complement a Rietveld refinement with DFT calculations, pair-distribution function modelling as well as external spectroscopic and other experimental methods.