Blind selected mapping (blind SLM) can effectively reduce the peak-to-average power ratio (PAPR) of both orthogonal frequency division multiplexing (OFDM) and single-carrier (SC) signals without side-information transmission. In typical blind SLM, maximum likelihood (ML) estimation is applied to find the de-mapping phase rotation sequence which gives the lowest Euclidean distance among all possible sequences, resulting in very high computational complexity. In this paper, we introduce a novel low-complexity 2-step estimation suitable for blind SLM. In the first step, the phase rotation sequence achieving the lowest Euclidean distance is searched by using the Viterbi algorithm. In the second step, verification and correction are carried out to choose a phase rotation sequence stored in the codebook, which has the lowest Hamming distance from the estimated sequence in the first step. It is confirmed by computer simulation that our proposed 2-step estimation achieves similar BER performance to the transmission without SLM and the transmission with blind SLM with the conventional ML estimation, but the proposed estimation technique requires much less complexity.