Background: Anaplastic lymphoma kinase (ALK) is one of the major oncogenic driver genes in lung cancer, and fusion as the main mutation type, of which EML4 is the most common fusion ligand. With the development of next-generation sequencing (NGS), some other rare fusion ligands and complex fusions have been discovered. The prognosis of patients with EML4-ALK fusion, non EML4-ALK fusion and multiple ALK fusion was different. Methods: We retrospectively analyzed NGS (including tissue, blood, or other body fluid samples) in 61 patients with ALK[+] lung cancer, including 59 patients with NSCLC and 2 patients with SCLC. Grouping was determined according to whether it was EML4-ALK fusion and multiple fusion. Results: In our data, the partner fused to the ALK gene are diverse, including EML4, KIF5B, TSN and others. Among them, EML4 was the most common ligand in ALK[+] NSCLC, accounting for 93.2% (55/59). Among the different fusion sites of EML4-ALK, V1 (E13:A20) and V3 (E6:A20) mutations were the most common, accounting for 45.5% (25/55) and 41.8% (23/55), respectively. Several novel ALK fusion partners were discovered in this study, including MCFD2, LINC01251, MVP17, CREG2, LINCO1498 and so on. In addition, the incidence of intergenic sequence-ALK fusions was 16.4% (10/61), such as ANXA4-ALK(A[intergenic]:A20), CREG2-ALK (C[intergenic]:A20), HCN1-ALK(H[intergenic]:A20), TSN-ALK(T[intergenic]:A20), RSAD2-ALK(R [intergenic]:A20) and so on. 19.0% (12/61) of the patient had multiple fusion, of which only one patient had multiple fusion of EML4-ALK (V2+V5'+V3a/b), and the remaining patients had co-fusion of EML4-ALK and non EML4-ALK (Table). V3 were more likely to coexist with non EML4-ALK fusions than V1 or other variants (P = 0.047). Multiple fusions were more likely to occur with non EML4-ALK fusions (P < 0.000). Compared with EML4-ALK fusion, mPFS treated with crizotinib in non EML4-ALK fusion were significantly shorter (18.2 vs. 8.5 months, P = 0.035). In the context of crizotinib treatment, multiple fusions were associated with worse mPFS compared with single fusions (13.8 vs. 8.4 months, P = 0.008). In addition, For SCLC patients with ALK fusion (n = 2), the PFS of crizotinib treatment were 8.0 and 30.6 months, respectively. Conclusions: Due to the heterogeneity of tumor development resulting in the diversity of ALK fusions, non EML4-ALK fusions and multiple fusions imply a poorer response to crizotinib. ALK gene status prior to ALK-TKI therapy helps predict drug efficacy and patient prognosis.