A new companding scheme for reducing the peakto-average power ratio (PAPR) of orthogonal frequency-division multiplexing (OFDM) systems is proposed in this study. It proceeds from speech signal processing similar to the earliest µlaw companding (MC) model. The proposed scheme compands (compresses and expands) the amplitudes of OFDM signals to a maximum of 1 − volt. Besides significantly reducing the PAPR, the proposed technique is also able to function as a limiter thus reducing the system complexity and limiting the amplitudes of OFDM symbols to a unity maximum voltage which does not exist in other companding PAPR techniques. Over frequencyselective fading channels with frequency domain equalization and using minimum mean square error (MMSE) to minimize the noise overhead, the proposed technique outperforms four other companding schemes over light and severe fading conditions. Lastly, we demonstrate that PAPR reduction using companding can dispense with correspondingly applying decompanding at the receiver as it amplifies the distortion noise thereby reducing the bit error ratio performance and increases the receiver complexity. We investigated the out-of-band interference of the proposed scheme and also show that it outperforms the other existing techniques by up to 5dB. Index Terms-Nonlinear companding, orthogonal frequency division multiplexing (OFDM), peak-to-average power ratio (PAPR), limiter, power amplifier (PA), out-of-band interference (OBI).