This study conducts a bond test of reinforced concrete confined by a new steel jacketing method without grouting, which was suggested previously. The steel jacket does not show composite behaviour with concrete because it has no grouting. The aims of this study are to determine the performance of the steel jacketing method in improving bond behaviour and to assess splitting stress developed during bond tests. For the purpose, this study uses concrete cylinders with the dimension of 100 mm × 200 mm and 150 mm × 300 mm (D × L), and D22 reinforcing bars (nominal diameter of 22·2 mm) are embedded in concrete. The specimens of 100 mm × 200 mm are designed to show splitting failure under unconfined conditions. On the contrary, the specimens of 150 mm × 300 mm are expected to produce pull-out failure under unconfined conditions. This study analyses the relation of circumferential strain to bond stress and suggests a model of splitting stress-strain for the confined concrete. The new steel jacketing method transfers the bond failure from splitting to pull-out mode and increases bond strength and toughness satisfactorily. This study suggests the ratio of bond stress to splitting stress as a function of slip that can be used to develop an analytical model of bond behaviour. In general, the bond strength increases with the confinement of the steel jackets; however, too heavy a confinement is not effective in continuously increasing the bond strength. In contrast to the bond strength, more confinement effectively decreases circumferential strain. The bond stress-circumferential strain curves generally show hook-shaped behaviour for pull-out failure mode for unconfined or confined specimens.