The distributed relay selection protocol recently developed by Bletsas et. al. provides a simple and practical mean of achieving cooperative diversity. However, for cooperative networks where relays employ the decode-and-forward (DF) method to send information, it cannot achieve full diversity unless each relay knows whether it decodes correctly or not. Moreover, the criterion used to select the best relay is not optimal since it does not minimize error probability. In this paper, we develop a distributed relay selection method for DF cooperative networks, that uses an error-probability minimizing criterion to select relay and can achieve full diversity without requiring the relay to know whether it decodes correctly. Moreover, we also develop a hybrid cooperative protocol that not only selects the best relay using an error-probability minimizing criterion but also chooses a better relaying method from DF and amplify-and-forward (AF). Our simulations show that our both protocols can achieve full diversity and outperform the protocol developed by Bletsas et. al.. Furthermore, our hybrid protocol offers better performance than the error-probability-minimizing DF protocol.