Accurate determination of anaplastic lymphoma kinase (ALK) rearrangements is critical in identifying ALK-positive patients for targeted therapy in non-small-cell lung cancer (NSCLC). Fluorescence in situ hybridization (FISH) is the current standard method to detect ALK rearrangements but is technically challenging and costly. We compared optimised immunohistochemistry (IHC), quantitative real-time polymerase chain reaction (qRT-PCR) and fluorescence in situ hybridization techniques in this study of 139 samples of advanced NSCLC with non-squamous histology. ALK alteration was found in 32.6 % (43/132) of patients by FISH, 32.9 % (45/137) of patients by IHC and 27.9 % (34/122) of samples by qRT-PCR (concordance rate of 96.9 % between FISH and IHC, 95.7 % between FISH and qRT-PCR, P < 0.001). IHC sensitivity and specificity were 97.7 % and 96.6 %, respectively, while the sensitivity and specificity of qRT-PCR were 89.2 % and 98.7 %, respectively. ALK rearrangements were more common in young patients (P = 0.007), non-smokers or light smokers (P = 0.008) and adenocarcinoma histology, especially with signet ring cell features (P < 0.001). Optimised IHC could be a useful method in screening ALK rearrangements in clinical practice with qRT-PCR as an alternative diagnostic tool to clarify specific ALK variants.