The binary-to-decimal (BTD) conversion is often used in computer.It is the key problem to calculate D first for converting the binary floating point number (BFPN) O.B₁B₂…B_n*2^B into the decimal 0.D₁D₂…D_m*10^D. In this paper，we comprehensively studied the nature of BFPNand inferred an important conclusion for calculating D precisely. Based on the conclusion we constructed a high efficient formula which can not only calculate D but also keep the total error minimum in application.