Recently, researches on space-time codes deploying multiple transmit and receive antennas have abruptly emerged due to its potential to greatly improve spectral efficiency. Particularly, space-time bit-interleaved coded modulation (ST-BICM) gives rise to very reliable bit error rate (BER) performance with quite flexible rate compatibility. In this paper, we propose an iterative linear minimum-mean-squared-error (LMMSE) channel estimation, detection, and decoding for ST-BICM systems over block fading channels in order to resolve more practical problems concerned with the performance degradation due to imperfect channel state information (CSI). The LMMSE channel estimation and detection are performed using a priori information fed back from the soft-input / soft-output (SISO) channel decoder in a computational efficient manner. The simulation results demonstrate that the performance of our proposed algorithm approaches to the matched filter bound (MFB) with an SNR loss of less than 1 dB at BER of 10-6, and its convergence of behavior is presented with the extrinsic information transfer (EXIT) charts.