The duration and magnitude of clinical response are unpredictable in ALK-rearranged non-small cell lung cancer (NSCLC) patients treated with crizotinib, although all patients invariably develop resistance. Here, we evaluated whether circulating tumor cells (CTC) with aberrant ALK-FISH patterns [ALK-rearrangement, ALK-copy number gain (ALK-CNG)] monitored on crizotinib could predict progression-free survival (PFS) in a cohort of ALK-rearranged patients. Thirty-nine ALK-rearranged NSCLC patients treated with crizotinib as first ALK inhibitor were recruited prospectively. Blood samples were collected at baseline and at an early time-point (2 months) on crizotinib. Aberrant ALK-FISH patterns were examined in CTCs using immunofluorescence staining combined with filter-adapted FISH after filtration enrichment. CTCs were classified into distinct subsets according to the presence of ALK-rearrangement and/or ALK-CNG signals. No significant association between baseline numbers of ALK-rearranged or ALK-CNG CTCs and PFS was observed. However, we observed a significant association between the decrease in CTC number with ALK-CNG on crizotinib and a longer PFS (likelihood ratio test, P ¼ 0.025). In multivariate analysis, the dynamic change of CTC with ALK-CNG was the strongest factor associated with PFS (HR, 4.485; 95% confidence interval, 1.543-13.030, P ¼ 0.006). Although not dominant, ALK-CNG has been reported to be one of the mechanisms of acquired resistance to crizotinib in tumor biopsies. Our results suggest that the dynamic change in the numbers of CTCs with ALK-CNG may be a predictive biomarker for crizotinib efficacy in ALK-rearranged NSCLC patients. Serial molecular analysis of CTC shows promise for realtime patient monitoring and clinical outcome prediction in this population. Cancer Res; 77(9); 2222-30. Ó2017 AACR.