Neural-symbolic integration is a field in which classical symbolic knowledge mechanisms are combined with neural networks. This is done to provide satisfactory computational capabilities from the network side and to exploit the descriptive power of symbolic reasoning. Logic Tensor Networks (LTNs) are a deep learning model that can be used to combine data with fuzzy logic to provide inferences and reasoning mechanisms over data. While LTNs have been shown effective in some contexts no detailed analysis on their capabilities for deductive logical reasoning has been conducted. In this paper we explore the capabilities and the limitations of LTNs in terms of deductive reasoning.