The histologic subtype of non-small cell lung carcinoma is important in selecting appropriate chemotherapy for patients with advanced disease. As many of these patients are not operative candidates, they are treated medically after biopsy for diagnosis. Inherent limitations of small biopsy samples can make distinguishing poorly differentiated lung adenocarcinoma (ADC) from squamous cell carcinoma (SCC) difficult. The value of histochemical and immunohistochemical markers to help separate poorly differentiated ADC from SCC in resection specimens is well established; however, the optimal use of markers in small tissue samples has only recently been examined and the correlation of marker expression in small tissue samples with histologic subtype determined on resection specimens has not been well documented. We address this issue by examining the expression of 9 markers (p63, TTF1, CK5/6, CK7, 34βE12, Napsin A, mucicarmine, NTRK1, and NTRK2) on 200 cases of ADC and 225 cases of SCC in tissue microarray format to mimic small tissue specimens. The single best marker to separate ADC from SCC is p63 (for SCC: sensitivity 84%, specificity 85%). Logistic regression analysis identifies p63, TTF1, CK5/6, CK7, Napsin A, and mucicarmine as the optimal panel to separate ADC from SCC. Reduction of the panel to p63, TTF1, CK5/6, and CK7 is marginally less effective but may be the best compromise when tissue is limited. We present an algorithm for the stepwise application of p63, TTF1, CK5/6, CK7, Napsin A, and mucicarmine in situations in which separation of ADC from SCC in small specimens cannot be accomplished by morphology alone.