Based on the well-known fact that social networks can provide effective mechanisms that help to increase the trust level between two trade partners, we apply a simple game-theoretical framework to derive transaction costs as a high risk of opportunistic behavior in a repeated trade relation determined by the density and size of trading networks. In the empirical part of the paper we apply a two stage procedure to estimate the impact of social network structures on farm’s transaction costs observed for different input and output markets. At a first stage we estimate a multiple input-multiple output stochastic Ray production function to generate relative shadow prices of three inputs and two outputs traded by farms. At a second stage a structural equation system is derived from the first order conditions of farm’s profit maximization to estimate simultaneously the of commodity specific transaction cost functions for all traded farm inputs and outputs. Estimation results based on a sample of 315 Polish farms imply a significant influence of social network structures on farm’s transaction costs. Moreover, estimated transaction costs correspond to a reasonable amount of farm specific shadow prices.