The growth cone is a highly motile tip structure that guides axonal elongation and directionality in differentiating neurons. Migrating immature neurons also exhibit a growth cone–like structure (GCLS) at the tip of the leading process. However, it remains unknown whether the GCLS in migrating immature neurons shares the morphological and molecular features of axonal growth cones and can thus be considered equivalent to them. Here, we describe a detailed method for time-lapse imaging and optical manipulation of growth cones using a super-resolution laser-scanning microscope. To observe growth cones in elongating axons and migrating neurons, embryonic cortical neurons and neonatal ventricular–subventricular zone (V-SVZ)-derived neurons, respectively, were transfected with plasmids encoding fluorescent protein–conjugated cytoskeletal probes and three-dimensionally cultured in Matrigel, which mimics the in vivo background. At 2–5 days in vitro, the morphology and dynamics of these growth cones and their associated cytoskeletal molecules were assessed by time-lapse super-resolution imaging. The use of photoswitchable cytoskeletal inhibitors, which can be reversibly and precisely controlled by laser illumination at two different wavelengths, revealed the spatiotemporal regulatory machinery and functional significance of growth cones in neuronal migration. Furthermore, machine learning–based methods enabled us to automatically segment growth cone morphology from elongating axons and the leading process. This protocol provides a cutting-edge methodology for studying the growth cone in developmental and regenerative neuroscience, being adaptable for various cell biology and imaging applications.