We describe the crystallization and structure determination of the 30 S ribosomal subunit from Thermus thermophilus. Previous reports of crystals that diffracted to 10 A resolution were used as a starting point to improve the quality of the diffraction. Eventually, ideas such as the addition of substrates or factors to eliminate conformational heterogeneity proved less important than attention to detail in yielding crystals that diffracted beyond 3 A resolution. Despite improvements in technology and methodology in the last decade, the structure determination of the 30 S subunit presented some very challenging technical problems because of the size of the asymmetric unit, crystal variability and sensitivity to radiation damage. Some steps that were useful for determination of the atomic structure were: the use of anomalous scattering from the LIII edges of osmium and lutetium to obtain the necessary phasing signal; the use of tunable, third-generation synchrotron sources to obtain data of reasonable quality at high resolution; collection of derivative data precisely about a mirror plane to preserve small anomalous differences between Bijvoet mates despite extensive radiation damage and multi-crystal scaling; the pre-screening of crystals to ensure quality, isomorphism and the efficient use of scarce third-generation synchrotron time; pre-incubation of crystals in cobalt hexaammine to ensure isomorphism with other derivatives; and finally, the placement of proteins whose structures had been previously solved in isolation, in conjunction with biochemical data on protein-RNA interactions, to map out the architecture of the 30 S subunit prior to the construction of a detailed atomic-resolution model.