Data reconciliation is an important part of quantum key distribution, which is also the key step for the continuous-variable quantum key distribution (CVQKD). Based on Leverrier′s prove about the security of CVQKD, a low density parity check code (LDPC) algorithm for multidimensional reconciliation is presented. The minimum convergence signal-to-noise ratio threshold of the algorithm is evaluated and the corresponding maximum key transmission distance of the CVQKD scheme is also estimated. By calculating the reconciliation efficiency and analyzing the noise, the maximum safe key capacity is further calculated. Simulation results indicate that the transmission distance between Alice and Bob increases from 30 km to 47 km, the decoding speed is 4 times of the sliced error correction (SEC) protocol, and the raw secret rate can reach 8.61 kb/s.