This paper proposes novel cooperative transmission protocols for delay-limited coherent fading channels with $N$ half-duplex and single-antenna nodes and one cell site. The protocols are evaluated using the Zheng-Tse diversity-multiplexing tradeoff. For the relay channel, two classes of cooperation schemes are investigated: Amplify and Forward (AF) and Decode and Forward (DF). An upper bound on the achievable diversity-multiplexing tradeoff is established for the AF class, and a new AF protocol, called Nonorthogonal Amplify and Forward (NAF), is proposed to achieve this bound. The NAF protocol is then extended to the general case with $(N-1)$ relays, outperforming the space-time coded protocol of Laneman and Wornell without requiring decoding/encoding at the relays. For the DF class, a dynamic decode and forward (DDF) protocol is developed that achieves the optimal tradeoff for multiplexing gains $0 \leq r \leq 1/N$. The DDF protocol is shown to dominate the AF protocols for all multiplexing gains and outperform the cooperative broadcast channel. In the cooperative multiple-access channel, a new AF protocol is proposed that achieves the optimal tradeoff for all multiplexing gains. The proposed protocols do not rely on orthogonal subspaces, allowing for more efficient resource utilization. The superiority of the DDF protocol is more significant in the cooperative broadcast channel, while the situation is reversed in the cooperative multiple-access channel. The results highlight the importance of exploiting the distributed nature of information to achieve optimal performance.This paper proposes novel cooperative transmission protocols for delay-limited coherent fading channels with $N$ half-duplex and single-antenna nodes and one cell site. The protocols are evaluated using the Zheng-Tse diversity-multiplexing tradeoff. For the relay channel, two classes of cooperation schemes are investigated: Amplify and Forward (AF) and Decode and Forward (DF). An upper bound on the achievable diversity-multiplexing tradeoff is established for the AF class, and a new AF protocol, called Nonorthogonal Amplify and Forward (NAF), is proposed to achieve this bound. The NAF protocol is then extended to the general case with $(N-1)$ relays, outperforming the space-time coded protocol of Laneman and Wornell without requiring decoding/encoding at the relays. For the DF class, a dynamic decode and forward (DDF) protocol is developed that achieves the optimal tradeoff for multiplexing gains $0 \leq r \leq 1/N$. The DDF protocol is shown to dominate the AF protocols for all multiplexing gains and outperform the cooperative broadcast channel. In the cooperative multiple-access channel, a new AF protocol is proposed that achieves the optimal tradeoff for all multiplexing gains. The proposed protocols do not rely on orthogonal subspaces, allowing for more efficient resource utilization. The superiority of the DDF protocol is more significant in the cooperative broadcast channel, while the situation is reversed in the cooperative multiple-access channel. The results highlight the importance of exploiting the distributed nature of information to achieve optimal performance.