To evaluate the expression of epidermal growth factor receptor (EGFR) mutation specific antibodies in invasive lung adenocarcinomas, and their sensitivity, specificity, as well as relationship to histological subtypes. Immunostaining with EGFR mutation-specific antibodies, del E746-A750 in exon 19 and L858R in exon 21, was performed in tissue microarrays of 884 cases of resection specimens to study the relationship between the immunophenotypes and morphologic subtypes. The sensitivity and specificity of the stains were compared with gene mutations detected by amplified refractory mutation system-polymerase chain reaction (ARMS-PCR). Of the 884 cases, the expression of del E746-A750 in exon 19 was 3+ , 2+ , 1+ and 0 in 7 cases (0.79%), 38 cases (4.30%), 129 cases (14.59%) and 710 cases (80.32%), respectively. For L858R in exon 21, 3+ , 2+ , 1+ and 0 staining were seen in 82 cases (9.28%), 93 cases (10.52%), 82 cases (9.28%) and 627 cases (70.93%), respectively. For both antibodies, positive expression (1+ or more) was mainly observed in lepidic, acinar and papillary predominant subtypes, and rarely seen in solid subtype or invasive mucinous adenocarcinoma (P=0.014 and 0.016). If 1+ to 3+ expression was set as positive, the specificity of exon 19/exon 21 reached 98.59%/92.98%, while the sensitivity was relatively lower (62.86%/88.89%). If 2+ to 3+ expression was read as positive, the specificity and sensitivity were 99.30%/97.37% and 25.71%/74.60% for exon 19/exon 21. If only 3+ expression was considered positive, the specificity was 100.0% for both antibodies, with a low sensitivity (8.57% for exon 19 and 34.92% for exon 21). Of the 18 cases with E746-A750 del in exon 19 based on molecular detection, the sensitivity of immunohistochemistry for exon 19 was 88.89% if a positive cutoff value ≥1+ was used; in contrast, of the 8 cases harboring other deletions in exon 19, only two cases were positive as 1+ . Both the EGFR mutation specific antibodies del E746-A750 in exon 19 and L858R in exon 21 demonstrate high specificity and relatively low sensitivity, and are mostly expressed in lepidic, acinar and papillary predominant subtypes, but rarely in solid subtype or invasive mucinous adenocarcinoma. For cases with 3+ expression, a mutational statue for EGFR is likely. For the 2+ positive cases, the accuracy to predict mutation almost reaches 90%, but molecular detection for confirmation is desirable. For the 1+ and negative cases, DNA-based test is essential to avoid false negativity.