With the increasing number of services and the massive data generated by the Web, a lot of services that have similar functionalities but different quality of service (QoS) attributes are candidates to fulfill the user’s requirements. The selection of the most appropriate services has therefore become even more difficult. To deal with this issue, a combination of several services through a composition process is carried out to form more sophisticated services. During the latest years, the QoS uncertainty has drawn a lot of interest due to the impact that may have on determining an optimal composite service, where each service is defined by several QoS properties (e.g., response time and cost), which are frequently changed due mainly to the environmental conditions. This paper focuses on the study of QoS-aware services composition approaches considering the uncertainty in QoS values based on a systematic literature review method. Unlike the existing literature reviews, this paper: (1) proposes a taxonomy of the uncertainty QoS-aware services composition approaches; (2) compares the studied approaches by taking into account the important criteria in the composition process such as the composition optimality, the time complexity and the techniques; (3) analyzes the studied papers and addresses a discussion while focusing on new challenges; and (4) provides the future research directions and perspectives.