We aimed to evaluate neuroendocrine pulmonary tumors (NEPT) by a novel method involving map tree construction by comparing all of the protein spots. We performed a proteomics analysis to assess the similarities in protein expression between neuroendocrine pulmonary tumors (NEPT), including typical carcinoids (TC), atypical carcinoids (AC), large cell neuroendocrine carcinomas (LCNEC) and small cell carcinomas (SCLC). Total protein lysates were obtained from seven histologically confirmed frozen NEPT tissues, including 1TC, 2 SCLC, and 4 cases ranging from AC to LCNEC. 2-DE demonstrated that TC was similar to normal lung. AC, LCNEC, and SCLC were similar to each other, forming a group separate from TC, however, SCLC at an early stage showed a similarity to TC. MALDI analysis detected 9 surrogate endpoint biomarkers, including eIF5A1, GST M3, cytokeratin 18 (CK 18), FK506-binding protein p59, p63, MAGE-D2, mitochondrial short-chain enoyl-coenzyme A hydratase 1, tranferrin and poly (rC) binding protein 1. Immunohistochemical staining revealed a gradual decrease in expression rate of p63 and CK 18 with poor differentiation of NEPT. Our results demonstrate that (1) the comparative proteomics of NEPT match the WHO classification except for AC and LCNEC; (2) SCLC show differences in their proteomics according to tumor stage; and (3) CK 18 and p63 may be useful as diagnostically and prognostically available markers.