Smartphones are widely used hubs of personal communication. With their many sensors, they are capable of monitoring social behaviours. Calls, messages, application usage and even face-to-face conversations can be captured by smartphones. These data are then used to study psychological, behavioural and clinical issues, thus providing an objective and continuous perspective on people’s social lives. However, compared with questionnaires, researchers are unfamiliar with the potential and challenges of smartphone social sensing. Moreover, previous studies utilised their own instruments and strategies, which caused repetitive work and threatened the generalisability or validity of results. To fill the gap between conceptual and empirical knowledge, we conducted a systematic review on empirical studies that applied passive smartphone social sensing. Forty-seven eligible articles were retrieved from 2,741 results from five major databases. A paradigm of reviewed studies was extracted, and the comprehensive procedures of passive smartphone social sensing, including applied sensors, experiment length and data analysis methods, were also summarised. This information can provide a general reference for researchers interested in applying passive smartphone social sensing. Benefits such as ubiquitousness, unobtrusiveness and personalisability and challenges such as privacy, accuracy and methodology were also reported. Moreover, fundamental research on reasonable sensor frequency choices, standardising features and implementing state-of-the-art technologies is recommended to enhance passive smartphone social sensing usability.