The diagnostic distinction between malignant pleural epithelial mesothelioma and pulmonary adenocarcinoma is often difficult and partially reliant on the use of immunohistochemistry (IHC). While there are several available IHC markers preferentially expressed in adenocarcinomas compared with mesotheliomas, there are no readily available or well-validated markers preferentially expressed in mesotheliomas and no markers are absolutely specific for either tumor. Thus, there always will be an element of doubt regarding the identity of lesions diagnosed by IHC using these markers and an undesirable reliance on negative results if the lesion is truly a mesothelioma. To help optimize the ability of currently available but imperfect markers to resolve adenocarcinoma and mesothelioma and to minimize the impact of false-negative results on their diagnosis, a systematic study was undertaken to identify multiple-marker immunostaining phenotypes that are the most specific and sensitive for each type of tumor. The study involved staining a series of malignant epithelial mesotheliomas (n = 34) and pulmonary adenocarcinomas (n = 103) with eight markers that were selected on the basis of previous reports of their relatively restricted specificities for one or the other of these lesions (and included carcinoembryonic antigen [CEA], tumor-associated glycoprotein 72 [B72.3], Leu-M1, vimentin, thrombomodulin, secretory component, carcinoma antigen-125, and periodic-acid-Schiff with diastase [for mucin]. Considering the markers one, two, and three at a time, all possible combinations of results were analyzed by computer to identify the phenotype most sensitive and specific for adenocarcinoma or mesothelioma. The best single marker was CEA (positive: 97% specific and sensitive for adenocarcinoma; negative: 97% specific and sensitive for mesothelioma). However, because examples of both tumor types expressed CEA, none of the study cases were unequivocally resolved and the risk of false-negative results was relatively high. The best two markers were CEA and B72.3 (both positive: 100% specific and 88% sensitive for adenocarcinoma; both negative: 99% specific and 97% sensitive for mesothelioma). The four possible combinations of results for these two markers resolved 68% of cases (93 of 137 cases), with less risk of false-negative results than a single marker. In general, several three-marker panels resulted in sensitivities and specificities similar to the two-marker panel of CEA and B72.3. In particular, the three-marker panel of CEA, B72.3, and Leu-M1 resulted in eight possible phenotypes that resolved 74% of cases (101 of 137 cases), with even further reduced risk of false-negative results.(ABSTRACT TRUNCATED AT 400 WORDS)