Summary The majority of human anti-tumour monoclonal antibodies (Mabs) isolated to date have been disappointing. Firstly, they react or cross react with intracellular cytoskeletal proteins or nuclear antigens and therefore are of limited value as blood borne agents. They are also generally of the IgM isotype and show relatively low intrinsic affinity for the primary epitope. Secondly, such Mabs can be generated from normal, non tumour bearing subjects at a frequency comparable to their production from tumour patients. This latter observation is true also for common autoantigens such as DNA and IgG since Mabs to these can also be generated from normal subjects in addition to autoimmune individuals. This article rationalises these observations in the context of the requirement for clinical use for human Mabs. It discusses the evidence that there is a potentially useful B cell response to be immortalised, and examines the consequences of the newly recognised phenomenon of monoclonal antibody multispecificity both on the methodology of their generation and on their subsequent use as imaging and therapeutic tools.