Trust is essential for most social and business networks in the web, and determining local trust values between two unfamiliar users is an important issue. However, many existing approaches to calculating these values have limitations in various constellations or network characteristics. We therefore propose an approach that interprets trust as probability and is able to estimate local trust values on large networks using a Monte Carlo simulation method. The estimation is based on existing indirect trust statements between two unfamiliar users. This approach is then extended to the SimTrust algorithm that incorporates both trust and distrust values. It is implemented and discussed in detail with examples. Our main contribution is a new approach which incorporates all available trust and distrust information in such a way that basic trust properties are satisfied.