Shoot growth and fruit production of 'Masui Dauphine' fig trees (Ficus carica L.) were compared between a novel training method and a control method, with various tree spacing. In the novel training, the shoots were elongated downward from a horizontal limb at 180 cm height, whereas control training had upward growing shoots from a horizontal limb at 40 cm height. Sprouted shoots of the novel training trees leafed a few days earlier than controls. The difference in training did not significantly affect longitudinal growth (the internode length and leaf area) of the shoots but, on the apical portion of the shoots, the shoot diameter and leaf weight per area in novel training were less than in controls. Many lateral shoots sprouted on the shoots of the novel training in autumn. The novel training prevented failure of fruit set, which was observed on the basal portion of control shoots with excess vigour owing to narrow tree spacing. The novel training promoted coloring of fruit on the basal portion of the shoots and depressed it on the apical portion. The size and weight of fruit tended to be reduced on the shoots that underwent novel training. The observed characteristics of novel training may be due to the change of lighting conditions and reduced photosynthetic rate due to downward shoot positioning.