Most current methods for analyzing the growth rate of plant seedlings are limited to low-throughput experimental configurations. We have developed an automatic system to investigate the dynamics of the growth of hypocotyls using Arabidopsis as model. This system is able to capture time-lapse infrared images of 24 seedlings automatically, with a spatial resolution of 2 μm per pixel and temporal interval of 5 min. Seedling length is rapidly calculated using automated geometric image-processing algorithms. With this high-throughput platform, we have investigated the genotype dependent difference of growth patterns, as well as the response to plant hormone - ethylene. Our analyses suggest that cytoskeleton function is not required in ethylene-induced hypocotyl inhibition. This novel integrative method can be applied to large-scale dynamic screening of plants, as well as any other image-based biological studies related to dynamic growth.