In this paper the formulas of asymptotic relative efficiency and asymptotic loss are derived for a rank binary integration detector after checking Capon’s regularity conditions. The results of calculation show that there exists an optimal rank quantization threshold which minimizes the asymptotic-loss. It is demonstrated that the optimal threshold is approximately equal to .8（N +1l）

where N is the number of reference cells. As N increases

the minimal asymptotic loss decreases and approaches to 0.94dB. Although the rank binary integration detector has a slighly higher asymptotic loss than that of a rank sum detector

it can be implemented more economically.