With only a few biocontrol products currently registered against Botrytis bunch rot (BBR) of grapes, there is a crucial need for new antagonistic strains that are able to survive and efficiently suppress B. cinerea under vineyard conditions. The aim of this study was to establish and follow a pre-selection process among potential antagonistic bacterial strains, previously identified in vivo for efficacy, and to carry out a further field screening assay using a reduced strain number. Ten bacterial strains were pre-selected and tested, in vitro and in vivo, to characterise their mode of action and population dynamics under simulated climatic regimes. Four candidate strains were then selected and characterised for high efficacy in vivo, known mode of action and marked survival ability. Some suitable additives for increasing strain adherence on grape berry surface were tested prior to field applications, indicating one commercial adjuvant for potential improved bacterial persistence in the field. The four strains were applied separately in an experimental Merlot vineyard near Bordeaux (SW France), either at five key phenological stages, or following a specially developed Disease Risk Index (DRI). The Bacillus ginsengihumi S38 strain treatments significantly reduced BBR incidence by 72-75% compared to the control, whereas sprays applied according to the DRI decision support system tended to improve disease control. The study validates a laboratory pre-selection process followed by a field screening step, resulting in a candidate B. ginsengihumi strain S38 with a high potential for BBR biocontrol and future development in vineyards.