Objective To evaluate the diagnostic value of serum neuron specific enolase (NSE) in patients with small cell lung cancer. Methods We searched MEDLINE, EMBASE, The Cochrane Library and other databases (1966 to March 2007) to collect studies which evaluated the diagnostic value of NSE in patients with small cell lung cancer. The heterogeneity of included studies was tested by the Cochrane Collaboration’s software RevMan 4.2. The Summary Receiver Operating Characteristic (SROC) curve and meta-analyses were performed by MetaDisc. Results Fifteen studies involving 4221 patients (672 SCLC and 3549 NSCLC patients, all diagnosed by the gold standard) were included. Meta-analyses showed that the heterogeneity among studies was high (P=0.000 2, I2=66.1%), the pooled sensitivity was 0.67 (95%CI 0.64 to 0.71) and the pooled specificity was 0.91 (95%CI 0.90 to 0.92). Subgroup analyses indicated that 4 of the studies which used the reagent supplied by The Academy of Military Medical Sciences (P=0.33, I2=13.4%, AUC= 0.9672, SE=0.0393) and another 4 which used the reagent supplied by Roche (P=0.23, I2=29.9%, AUC=0.8311, SE=0.0836) had no heterogeneity. Conclusion NSE could be regarded as one of the reference tests in patients with small cell lung cancer, but more high quality trials are required.