Summary A systematic review and meta-analysis was done on the use of PCR tests for the diagnosis of invasive aspergillosis. Data from more than 10 000 blood, serum, or plasma samples obtained from 1618 patients at risk for invasive aspergillosis were retrieved from 16 studies. Overall, the mean diagnostic odds ratios (DORs) of PCR for proven and probable cases were similar whether two consecutive positive samples were required to define positivity (DOR 15·97 95% CI 6·83–37·34) or a single positive PCR test was required (DOR 16·41 95% CI 6·43–41·88). Sensitivity and specificity of PCR for two consecutive positive samples were 0·75 (95% CI 0·54–0·88) and 0·87 (95% CI 0·78–0·93), respectively, and if only a single positive sample was required, these values were 0·88 (95% CI 0·75–0·94) and 0·75 (95% CI 0·63–0·84), respectively. Whereas specificity based on a single positive test was significantly lower (p=0·027) than two positive tests, the sensitivity and DOR did not differ significantly. A single PCR-negative result is thus sufficient to exclude a diagnosis of proven or probable invasive aspergillosis. However, two positive tests are required to confirm the diagnosis because the specificity is higher than that attained from a single positive test. Populations at risk varied and there was a lack of homogeneity of the PCR methods used. Efforts are underway to devise a standard for Aspergillus sp PCR for screening, which will help enable formal validation of PCR and estimate its use in patients most likely to benefit.