Gully erosion is a severe form of soil erosion. However, current understanding of the relative dominance of processes that drive the development of gully erosion is limited. This study investigated the development of gully erosion, and the contribution of mass movement and water erosion to gully erosion based on fourteen scouring experiments on two field plots and terrestrial laser scanning (TLS). A three-dimensional (3D) method was proposed to derive the volume of gully erosion based on point clouds with topographic change information. The results showed that TLS-derived gully sediment yields were in a high agreement with the measured sediment yield (R2 > 0.85, p < 0.01, and mean relative/absolute error < 11.38%/5.47 kg), which demonstrated the reliability of TLS and the newly proposed 3D volumetric calculation method in the gully erosion monitoring. We found that the development of gully erosion consisted of horizontal and vertical cycles. The horizontal development of gullies was dominated by a cycle of gully head retreat that was mainly driven by water erosion as well as gully head expansion that was considerably driven by collapses. The vertical development was driven by the co-evolution of collapses and caves that enhance the development of one another. Our experiments showed that mass movements were generally not a major source of gully erosion mass (1.09%–50.80%) or sediment yield (0.90%–24.71%), although they considerably contributed to the morphological development of gully heads (e.g. 26.74% to 96.09% of the increase in the maximum length of gully head and 39.03% to 94.80% of the change in gully head area). We also found that mass movements significantly impacted gully erosion mass (p < 0.05), while their relationships with sediment yield tended to be complex and depended on the impact of mass movements on the broader runoff and erosion processes. Besides, the extremely high collapses were also found to heavily impact the relationship between the collapse mass and the development of gully erosion, implying that further experiments are needed in the future to enrich the observations and thus improve the above-mentioned understanding. We conclude that by using TLS data in an experimental set-up, gully erosion processes can be captured with fine detail. This provides new insights into the process-understanding of gully erosion.