Loneliness is the feeling associated with a perceived lack of qualitative and quantitative aspects of social relationships. Loneliness is thus evidently intwined with individuals’ social behaviors in day-to-day life. Yet, little is known about the bidirectional pathways between loneliness and social interactions in daily life. In this study, we thus investigate (a) how loneliness predicts the frequency and duration of social interactions and (b) how frequency and duration of social interactions predict changes in loneliness. We examine these questions using fine-grained ambulatory-assessed sensor data of student’s social behavior covering 10 weeks (N_participants = 45, N_observations = 74,645). Before (T1) and after (T2) the ambulatory assessment phase, participants completed the UCLA loneliness scale, covering subscales on intimate, relational, and collective loneliness. Using multistate survival models, we show that T1 loneliness subscales are not significantly associated with differences in social interaction frequency and duration– only relational loneliness predicted shorter social interaction encounters. In predicting changes in loneliness subscales (T1-T2), only the mean duration of social interactions was negatively associated with collective loneliness. Thus, effects of loneliness on the structure of social interactions may be small or limited to specific forms of loneliness, implying that the quality of interactions may be more important.