The methods involved in determining the 850 kDa structure of the 30S ribosomal subunit from Thermus thermophilus were in many ways identical to those that are generally used in standard protein crystallography. This paper reviews and analyses the methods that can be used in phasing such large structures and shows that the anomalous signal collected from heavy-atom compounds bound to the RNA is both necessary and sufficient for ab initio structure determination at high resolution. In addition, measures to counter problems with non-isomorphism and radiation decay are described.