In reinforced concrete (RC) structures, bond behavior between concrete and deformed steel bar (rebar) has a great effect on the structural behavior of members including voided slabs. According to previous research, bond behavior of a rebar in voided slabs is different from that in general RC slabs by the concrete inner cover formed between rebar and void. However, the bond characteristics of rebar in voided slabs have not been clearly verified yet. This study investigated the bond characteristics of rebar in the donut-type voided slab. Like other voided slabs, the donut-type void-shaper make concrete inner cover formed between rebar and void, and some part of the donut-type voided slab has smaller inner cover thickness than 2.5 db. Furthermore, inner cover thickness changes along the longitudinal rebar due to the shape of void. These were expected to affect the bond behavior of rebar in the donut-type voided slab. In this study, pull-out tests were performed to find out the effect of bond condition by the region due to voids. To this end, the donut-type voided slab was divided into three regions according to the shape of void, such as ‘Insufficient region’, ‘Transition region’, and ‘Sufficient region’. Main variables were inner cover thickness, embedded length, and bond location, which is affected by void shape. Bond characteristics of rebar in the donut-type voided slab were evaluated through comparison of bond stress–slip relationship, maximum bond strength, and bond stress distribution of each region. Test results showed that the average bond stress was found to be 25~50% lower than that of the general RC slab in the region of the inner cover thickness of 2.5 db, which was not satisfied due to void, and the bond stress–slip relationship and the bond stress distribution of the donut-type voided slab was also different from that of the general RC slab according to the type of bond region and inner cover thickness. In addition, the calculation method of bond strength of rebar in donut-type voided slab was proposed, which indicated that the error range between the estimated value and the experimental value was within 5%.