One of the advantages of a Conversational Agent is that it stimulates social presence among Distance Education students, providing continuous psychological and pedagogical support. However, the interaction between students and virtual agents is often functionally ineffective. In this regard, this study aims to investigate students' mood states, inferred by chat log analysis of interactions between students and a Conversational Agent, relating such mood states with the students' perceptions about the tool. The study was conducted with a sample of students attending a Distance Education post-graduation course. The results showed the relationship between interest, utility and satisfaction perceived by the students with mood states detected by the log analysis. Also, the underutilization of the Conversational Agents was diagnosed, and improvements to be implemented so that negative mood states which can be overcome were identified. The research brings contributions on building better Conversational Agents for educational purposes.