Abstract: Increasing the number of iteration steps can decrease BER (bit error rate) but increase time-delay. A compromise is reached between the errors and time-delay, and an adaptive iterative decoding algorithm is improved through an investigation on the principle of turbo codes and iterative decoding technique, in which a novel strategy is introduced to determine when to terminate the iterative process. Using the tail bits generated from decoder to detect errors and designing properly the minimum number of iteration steps can efficiently reduce the average number of iteration steps with low BER. Computer simulation showed that the improved algorithm can efficiently decrease the time-delay.