Turbo code, created in 1993, has a good performance in itself . Many researchers have made efforts at finding the encoding code and interleaver for better performance and diminishing the complexity and the long-time delay of the decoder which are weak points of Turbo code, leading to effective hardware implementation. The BER curve of Turbo code consists of two regions, error-floor region and waterfall region. In the error-floor region, the codewords which have the minimum input weight are the main factors on error performance. In the waterfall region, the code multiplicity as well as the minimum-input-weight codewords is the main factor on error performance. In this paper, information-parity-inserted Turbo code is introduced. By inserting parity bits in the information bits, the minimum input weight can be changed to a larger number and error performance in the error-floor region can be improved. Moreover in the decoding process, the iteration number is changed adaptively according to the information parity check, so that sufficiently error-free decoding can be performed.