Abstract: Aimed at the searching for good codes in the Turbo code and serially concatenated convolutional code, the paper proposes a novel trellis unity project to calculate the weight distribution of convolutional code. Based on the weight distribution produced by encoding operation in the forepart of trellis, the new project makes iterative trellis unity by doubling previous trellis path length every time to get the whole weight distribution. Under a middle or low interleaving delay, the trellis unity project has a similar calculation complexity with the conventional method which needs to utilize state transference figure. But the new project can omit much preparative calculation work, so it is of significance for practical work.