In clinical practice, the finding of an elevated mean corpuscular volume (MCV), macrocytic anaemia or specific neurological symptoms is often the reason to test for vitamin B12 (B12) deficiency. Use of the MCV as a test for the detection or exclusion of B12 deficiency is only justified if the diagnostic accuracy is sufficiently high. However, the sensitivity and specificity are not well known. We performed a systematic review of the diagnostic value of an elevated MCV for B12 deficiency in both anaemic and non-anaemic patients. Of approximately 3500 titles and/or abstracts that were screened, 37 original papers contained usable data. The population under study proved to be the characteristic of major influence on the study outcome. Pooling of data from different studies was performed in subsets of the data corresponding to the different populations studied. The cut-off levels of both MCV and serum B12 had a significant influence on the study outcomes. The data, however, were pooled without taking these cut-off levels into account. The pooled estimates should be interpreted with this limitation in mind. The reference standards were (1) a low serum B12 concentration and (2) a B12 deficiency confirmed by low serum B12 combined with additional diagnostic investigations. In the population that was randomly screened for low serum B12, the sensitivity of the MCV for B12 deficiency was 17%, whereas the sensitivity was 30% for B12 deficiency in patients with anaemia. When measurement of serum B12 was ordered to exclude B12 deficiency as part of the patients' treatment, the sensitivity was 30% for low serum B12 concentration, 58% for B12 deficiency and 75% for B12 deficiency in patients with anaemia. In the population with pernicious anaemia, the sensitivity was far from perfect (77%). In the five studies that reported data on the positive predictive value of the MCV for B12 deficiency, this ranged from 0% (0/6) to 55% (11/20). This systematic review shows that a considerable number of B12-deficient patients will remain unnoticed when the MCV is used to rule in patients for further evaluation. Depending on the population studied, up to 84% of cases will than be missed. The MCV can be used to make the diagnosis of B12 deficiency more--or less--probable. An elevated MCV justifies the measurement of serum B12. The MCV should not be used as the only parameter ruling out the diagnosis of B12 deficiency.