An estimate-and-forward(EF) scheme for single-input single-output(SISO) and multiple-input multiple-output(MIMO) full-duplex two-way relay networks is proposed and analyzed. The relay estimates the received signal from two terminal nodes by a minimum mean squared error(MMSE) estimation and forwards a scaled version of the MMSE estimate to the destination. The proposed EF outperforms conventional amplify-and-forward(AF) and decode-and-forward(DF) across all signal-to-noise ratio(SNR) region. Because its computational complexity is high for relays with a large number of antennas(large MIMO) and/or high order constellations, an approximate EF scheme, called list EF, are thus proposed to reduce the computational complexity. The proposed list EF computes a candidate list for the MMSE estimate by using a sphere decoder, and it approaches the performance of the exact EF relay at a negligible performance loss. The proposed forwarding approach also could be used to other relay networks, such as half-duplex, one-way or massive MIMO relay networks. An estimate-and-forward (EF) scheme for single-input single-output (SISO) and multiple-input multiple-output (MIMO) full-duplex two-way relay networks is proposed and analyzed. The proposed EF outperforms conventional amplify-and-forward (AF) and decode-and-forward (DF) across all Because its computational complexity is high for relays with a large number of antennas (large MIMO) and / or high order constellations, an approximate EF scheme, called list EF, are thus to reduce the computational complexity. The proposed list EF computes a candidate list for the MMSE estimate by using a sphere decoder, and it approaches the performance of the exact EF relay at a negligible performance loss. The proposed forwarding approach also could be be used to other relay networks , such as half-duplex, one-way or massive MIMO relay networks.