ABSTRACT : We performed a meta-analysis of diagnostic data to evaluate the performance of Histoplasma antigen detection tests for diagnosing histoplasmosis. We included all studies involving human subjects that assessed the performance of any antigen detection test for histoplasmosis in urine or serum by carrying out an exhaustive and reproducible search of the literature between 1980 and 2014 from four databases. Quality of the articles was assessed, and meta-analysis was performed under the random effects model, calculating sensitivity, specificity, likelihood and odds ratios, and ROC curve using Meta-DiSc(es). Nine out of a total of 23 studies met strict quality criteria and were therefore included. The overall sensitivity for antigen detection in serum and urine was 81 % (95 % CI 78–83 %), while specificity was 99 % (95 % CI 98–99 %). Sensitivity for antigenuria and antigenemia was 79 % (95 % CI 76–82 %) and 82 % (95 % CI 79–85 %), respectively; specificity values were 99 % (95 % CI 98–100 %) in urine and 97 % (95 % CI 96–98 %) in serum. The positive and negative likelihood ratios were 49.5 (95 % CI 20.7–118.7) and 0.19 (95 % CI 0.14–0.26), respectively, while the diagnostic OR was 362 (95 % CI 121.2–1080.3) and area under the curve was 0.99. In conclusion, the performance of Histoplasma antigen detection assay of urine was not significantly different from that of blood, indicating that antigenuria and antigenemia have equal diagnostic value in histoplasmosis.