To assess the justification for the routine use of investigations in the diagnosis of bacterial causes of sore throat. The literature from 1945 to 1990 was systematically screened to identify studies that addressed diagnosis of bacterial infection and the efficacy of antibiotics in sore throat, using the key-words "pharyngitis" and "tonsillitis". Difficulties were identified with clinical methods and investigations that identify streptococcal infections. The practice of throat-swab culture--the "gold standard"--appears to have developed as a strategy to protect patients from acute rheumatic fever. However, this method may be limited in its usefulness for protection against acute rheumatic fever because: (i) in many cases in which the streptococcus is isolated from symptomatic patients there is no serological evidence of infection; (ii) there are very high asymptomatic carrier rates of the streptococcus; (iii) even after adequate treatment with penicillin there are high bacteriological failure rates; and (iv) those organisms that can be isolated from the mucosal surface are a poor representation of organisms lying deep in the tissues. Evaluation of other diagnostic techniques such as Gram's stain and rapid antigen testing, as well as decision analysis, has also been hampered by the difficulties encountered with use of this inadequate gold standard. There is little indication from the literature that any routine system of identifying bacterial causes of sore throat is helpful to the clinician.