The authors evaluated methods of recognition of neuroendocrine differentiation in lung cancer because this is thought to bear prognostic value. One hundred forty lung tumors were divided by immunohistochemical analysis using neuroendocrine markers (neuron-specific enolase, Leu7, chromogranin, neural cell adhesion molecules, SL11/14, and Leu 19) into two groups of 51 neuroendocrine tumors positive for three neuroendocrine markers and 89 non-neuroendocrine tumors. Biochemical determination of enolase activity and isoenzyme distribution showed that the level of total enolase activity was similar between neuroendocrine and non-neuroendocrine tumors, αγ and γγ enolase isoenzyme percentages were higher in neuroendocrine tumors. A cut-off of γ enolase % (αγ/2 + γγ) at 14% gave a sensitivity rate of 84% and specificity rate of 97% in separating neuroendocrine and non-neuroendocrine tumors, whereas immunohistochemical analysis using anti-γ enolase had low specificity (68%) and immunohistochemical analysis with Leu 7 and chromogranin had high specificity (97%) and low sensitivity (37% and 60%) levels for neuroendocrine tumors. The best prediction of neuroendocrine differentiation was obtained using immunohistochemical analysis against neural cell adhesion molecules combined with biochemical estimation of enolase using γ enolase of 14% and a γγ isoenzyme more than 3%.