New technologies are revolutionizing diagnostic microbiology, but implementation of methodological advances in test evaluation has been extraordinarily slow. Published reports frequently fail to clearly distinguish between studies of analytical accuracy and those of clinical diagnostic performance. We argue that the evaluation of sensitivity and specificity, while necessary and occasionally problematic, is often inadequate to define the appropriate role of a new diagnostic test. To determine whether a test adds additional (incremental) value to the diagnostic information already available to the clinician, evaluation studies with a multivariable approach may also be needed. The impact of a test on patient outcome is best measured by a randomized controlled trial, but this may be unnecessary in certain well-defined situations. To improve the quality of published test evaluations close collaboration between clinicians, clinical microbiologists, and epidemiologists, as well as insistence by journal editors on the use of established reporting standards are essential.