: The rapid growth of communication technologies in the past years has enabled development of collaborative peertopeer applications. The reliable col laborative application must implement some methods which help to detect mali cious entities and avoid collaboration with them. Such systems are based on build ing trust relationships between entities; their design is difficult due to a decentral ized architecture and general distrust among unknown entities. In this paper we analyze requirements for reliable trust management in an insecure environment like Internet. We propose criteria which reliable trust management should meet and we compare several published systems according to these criteria. We also propose guidance on design and evaluation of trust management techniques.