Vertical climbing is a common movement in industrial field, which requires coordination of entire body joints and is a high-risk task for workers. Exoskeletons have the ability to assist wear to perform vertical climbing and reduce the risk of failed motion. However, raw studies have paid attention on exoskeleton-assisted vertical climbing. In order to develop vertical climbing assistance strategy for exoskeleton system, it is essential to obtain the inherent control strategy of human subject for this movement. Muscle activations, which can be observed by surface Electromyography (sEMG) signals, are able to present muscle activity or human potential control intent. Therefore, we adopt muscle synergy technology to analyze human vertical climbing movement. Firstly, the whole process is divided into four phases according to the movement data and sEMG data collected in experiment. Then, the preprocessed signals were decomposed using a non-negative matrix factorization (NNMF) algorithm. Finally, four muscle synergistic responses are extracted by variance accounted for (VAF) approach. The results of the experiments showed that the subjects’ muscle synergy between the different phases was not highly correlated. The final finding may indicate that during vertical climbing, human beings change muscle activation patterns to adapt to the impacts from body stability and to realize vertical climbing.