Superposed multicarrier transmission scheme is known to improve frequency utilization efficiency when several wireless systems share the same spectrum. To suppress the effect of interference, forward error correction (FEC) metric masking is proposed. In the technique, log-likelihood ratio (LLR) that corresponds to superposed band is set to zero, because received bits that correspond to superposed band is unreliable. However, to apply FEC metric masking, the information about superposed band must be known at the receiver beforehand. Furthermore, the received bits contain channel estimation error, which is a cause of degradation. In this paper, we propose an iterative estimation technique of power of undesired signal (noise, interference, and channel estimation error) for superposed multicarrier transmission. We use the estimated power of the undesired signal to calculate the LLR that takes the channel estimation error into account, since including this extra information about the channel improves the BER. The proposed scheme estimates the power of undesired signal on each subcarrier, thus the information about superposed band is not required. Simulation results show that the accuracy of estimating undesired signal becomes more reliable as the number of estimation increases, so that BER becomes better as a result of iterative estimation.