Small cell lung cancer (SCLC) remains as the most lethal form of lung cancer with very poor prognosis. High metastatic capability and strong plasticity is the hallmark of SCLC. However, the exact cell subpopulation responsible for SCLC metastasis and critical regulators in contribution to SCLC plasticity still remain largely unclear. Growing evidence has revealed the importance of epigenetic reprograming during cancer malignant progression. Mammalian SWI/SNF complex BAF is an adenosine triphosphate (ATP)-dependent multi-component machinery, essential for remodeling chromatin architecture. Smarca4 (Brg1), the core ATPase component of BAF complex is preferentially required for the progression of SCLC with MAX (Myc-associated factor) inactivation. Also, it still remains elusive about the exact function of Brg1 and SWI/SNF complex in SCLC metastasis. Using the well-established Rb1L/L/Trp53 L/L (RP) mouse model, we here identify the NCAMhiCD44lo/- subpopulation derived from RP model as SCLC metastasizing cells (SMC). Moreover, we find that the SMC is progressively transitioned from the NCAMloCD44hi cells (Non-SCLC metastasizing cells, Non-SMC) during malignant progression of SCLC. Through combined analysis of ATAC-seq and RNA-seq data of SMC vs. non-SMC, we observe a dramatic alteration in chromatin accessibility between these two subpopulations, and TAZ, the core transcription cofactor of the Hippo pathway, is epigenetically silenced by the SWI/SNF complex during SCLC malignant progression. Knockout of Brg1 in RP mouse promotes TAZ expression and abrogates the phenotype transition from Non-SMC to SMC and almost completely blocks SCLC metastasis. Allograft assay further confirms TAZ as the critical molecular switch during the phenotypic transition and SCLC metastasis. In this study, we identify the NCAMhiCD44lo/- subpopulation derived from RP model as SMC subpopulation, which is progressively transitioned from the Non-SMC during SCLC malignant progression. Through combined analysis of ATAC-seq and RNA-seq data, we observe a dramatic alteration in chromatin accessibility between these two subpopulations, which points to the alteration of SWI/SNF complex activation. We further find that genetic disruption of the SWI/SNF chromatin-remodeling complex through Brg1 knockout abrogates the phenotype transition from Non-SMC to SMC and blocks SCLC metastasis. Mechanistically, we find that TAZ, the core transcription cofactor of the Hippo pathway, is epigenetically silenced by the SWI/SNF complex. Down-regulation of TAZ promotes the transition from Non-SMC to SMC and facilitates SCLC metastasis. Conversely, ectopic TAZ expression or TAZ activation by digitoxin treatment reversibly promotes the transition from SMC to Non-SMC and thus alleviates SCLC metastasis. Importantly, the treatment of digitoxin, a clinically used drug for heart failure treatment, promotes TAZ activation and inhibits SCLC phenotypic transition and metastasis. Through the immunostaining analyses of 101 Chinese SCLC specimens, we further find that low TAZ expression significantly correlates with SMC features and poor patient prognosis. Collectively, these data uncover an important subpopulation in contribution to SCLC metastasis and highlights TAZ as the critical molecular switch in this process. Moreover, TAZ activation by digitoxin treatment provides an effective avenue to control SCLC phenotypic transition and metastasis, providing a potential therapeutic avenue for the metastatic SCLC management in clinic.