The well-known classification of neuroendocrine neoplasms of the lung into four major subtypes (including typical and atypical carcinoids and small- and large-cell neuroendocrine carcinomas) has a proven prognostic validity but only partially helps to predict the response to specific therapies. Therapeutic biomarkers are incompletely known and include morphological, immunophenotypic, and molecular markers. Morphology alone has no specific predictive role, nor has any immunophenotypic marker been proven to bear predictive implications. Ki67 is a relevant prognostic marker and can indirectly predict response to chemotherapy, when levels are extremely high in high-grade neuroendocrine (NE) carcinomas. The expression of somatostatin receptors, especially of the type 2A, has been shown to predict response to somatostatin analog treatments, paralleling the information derived from octreotide scintigraphy. mTOR pathway is targeted by specific inhibitors, but the exact cellular molecules predicting response are still to be defined. It seems that high levels of phosphorylated forms of mTOR and of its downstream factor S6K are associated to a better response to rapalogs in experimental models. Data from gene expression profiling and mutational analyses are currently emerging, providing a more detailed map of different molecular activation pathways, potentially leading to a more accurate molecular classification of lung NE tumors as well as to the discovery of new therapeutic targets. The combination of mutational profiles with those of upregulated or downregulated genes also by gene gains or losses may ultimately provide a better characterization of NE tumor histological types in terms of response to specific chemotherapy or biotherapy.