Kimberlite magmas, the primary source of diamonds, have many features indicative of explosive eruptions and high volatile contents. The main approaches used to establish exsolution of fluid during magma ascent include theoretical modeling and experimental estimates of volatile solubility in kimberlite-like melts. Both approaches are hampered by the poorly constrained composition of kimberlite melts. Resorption features on diamonds are very sensitive to the presence and composition of the kimberlite fluid as well as to temperature and pressure. Here, we use direct evidence from diamond resorption features as a new method for investigating the parameters of fluid exsolution. The method is based on experimental reproduction of diamond resorption in kimberlite melts with and without an exsolved fluid phase. We studied 802 diamonds from two kimberlites (BK1 and AK15) from the Orapa cluster, Botswana. Samples from the BK1 pipe include three lithologies: two coherent kimberlites (CK-A and CK-B) and a pyroclastic kimberlite (massive volcaniclastic kimberlite, MVK). The known depth of diamond samples in each kimberlite lithology allows us to demonstrate an increase in the intensity of kimberlite-induced resorption with depth of diamond recovery in the drill holes. Each kimberlite lithology has a different proportion of diamonds with kimberlite-induced resorption, which is unique in style in each lithology: glossy surfaces in MVK due to reaction with C-O-H fluid, rough corroded surfaces in CK-B due to reaction with volatile-undersaturated melt, and a combination of glossy surfaces with corroded features in CK-A due to an overprint of melt resorption after fluid resorption. Both diamond resorption and kimberlite textures in the BK1 kimberlite show evidence of fluid exsolution only in CK-A and MVK lithologies, but no fluid presence in CK-B. The observed diamond resorption features may be controlled by (1) a temporary separation of the rising magma column into a bubblerich head and bubble-poor volatile-depleted tail and (2) fluid exsolution at depths greater than decompressional degassing. We discuss how the depth of fluid exsolution from kimberlite melt may affect the diamond grade and the resorption of diamond populations in a kimberlite.