This paper proposes an iterative low complexity soft input decoding with assistance of the lattice reduction for overloaded MIMO. The proposed soft input decoding applies two types of lattice reduction-aided linear filters to estimate log-likelihood ratio (LLR) in order to reduce the computational complexity. A lattice reduction-aided linear with whitening filter is introduced for the LLR estimation in the proposed decoding. The equivalent noise caused by the linear filter is mitigated with the decoder output streams and the LLR is re-estimated after the equivalent noise mitigation. Furthermore, LLR clipping is introduced in the proposed decoding to avoid the performance degradation due to the incorrect LLRs. The proposed decoding achieves about 2dB better BER performance than the soft decoding with the exhaustive search algorithm, so called the MLD, at the BER of 1.0E-4, even though the complexity of the proposed decoding is 1/10 as small as that of the soft decoding with the exhaustive search.