Data from beam-based bond tests for FRP bars in the literature were collected and regression analyses were conducted for the data of splitting failure. Average bond strengths obtained from splice tests were found to be lower and more affected by C/$d_b$ values than average bond strengths from anchorage tests, indicating needs of new design equation for the splice length of FRP bars based on the data of splice tests only. In addition, the variation of bond strengths was greater than that of tensile strengths of FRP bars and, therefore, a new safety factor should be involved for the design equation. Five percent fractile coefficients were used to develop the design equations based on the assumption that load and resistance factors for FRP reinforced concrete structures are same to the factors for steel reinforced concrete structures. The proposed design equations give economical and reliable lengths for development and splice of FRP bars. The proposed equation for splice provides shorter lengths than the ACI 440 equation in case of C/$d_b$ of 3.0 or greater. Because FRP bars are expected to be used in slabs and walls exposed to weather with thick cover and large spacing between bars, the proposed equation gives optimal splice lengths.