Affinity maturation is a process by which low-affinity antibodies are transformed into highly specific antibodies in germinal centres. This process occurs by hypermutation of immunoglobulin heavy chain variable (IgH V) region genes followed by selection for high-affinity variants. It has been proposed that statistical tests can identify affinity maturation and antigen selection by analysing the frequency of replacement and silent mutations in the complementarity determining regions (CDRs) that contact antigen and the framework regions (FRs) that encode structural integrity. In this study three different methods that have been proposed for detecting selection: the binomial test, the multinomial test and the focused binomial test, have been assessed for their reliability and ability to detect selection in human IgH V genes. We observe first that no statistical test is able to identify selection in the CDR antigen-binding sites, second that tests can reliably detect selection in the FR and third that antibodies from nasal biopsies from patients with Wegener's granulomatosis and pathogenic antibodies from systemic lupus erythematosus do not appear to be as stringently selected for structural integrity as other groups of functional sequences.