The most successful method to date for predicting protein tertiary structure from primary sequence data is homology modelling based on alignment with similar sequences of known structure. The use of a variety of computing methods to identify the best similarities is discussed. Model building based on alignments and the construction of libraries of side chain conformations is described. The application of sequence alignment modelling to the structure prediction of EcoKI type I DNA methyltransferase is shown in the context of corroborative laboratory experiments. Finally, a method is presented which incorporates sequence alignment with secondary structure prediction. A program - sss_align - which incorporates this method, was used to make blind fold recognition predictions as part of an international collaborative exercise in the critical assessment of methods of protein structure prediction ('CASP2'). It was shown by this and other assessment methods that sss_align will detect similarities between sequences which exhibit as little as 15% identity.