In the last years wireless communications have experienced a fast growth due to the high mobility that they allow. However, wireless channels have some disadvantages, like multipath fading, that make them difficult to deal with. A modulation that efficiently deals with selective fading channels is OFDM. The advancements in digital signal processing and very large scale integrated circuits allow efficient and cost-effective implementation of the FFT operations making OFDM an attractive solution for wireless channels. An important disadvantage of OFDM systems is their high peak-to-average power ratio (PAPR). High PAPR degrades OFDM signals by forcing the analog amplifier to work in the nonlinear region, distorting this way the signal and making the amplifier to consume more power. There are several ways to reduce the PAPR of OFDM signals which can be divided in two groups: the techniques that non-linearly distort the signal and the ones that reduce the PAPR without distorting the signal. A simple technique, belonging to the first group, is to clip the signal to a maximumallowed value, at the cost of BER degradation and out-of-band radiation. Clipping does not add extra information to the signal and high peaks occur with low probability so the signal is seldom distorted. Out-of-band radiation can be reduced by filtering at the transmitter, the filter used in this project consists on a FFT-IFFT pair which is easier to implement than traditional FIR filters and allows the implementation of the clip&filter set several times in order to reduce the peak regrowth that filtering introduces. The BER degradation can be mitigated by reconstructing the signal at the receiver. We analyzed the performance of some commonly used techniques that iteratively try to guess the original symbols and proposed an improvement for one of those techniques. The methods have been tested with Matlab and from the simulation results some conclusions are extracted.