Affective computing focuses on technologies and theories that advance understanding of human affect, considering emotion and cognition in the design of related technologies to fulfill human needs, which gains substantial attention of researchers all over the world. To provide an insight into affective computing researches, this paper utilizes the method of bibliometric analysis to obtain information with respect to when and where the researches were performed by whom and how the mainstream contents evolved over the years. BibExcel and CiteSpace were employed to conduct the performance analysis and co-citation network analysis, including the analysis of the performance of countries, journals, institutes, authors and research hotspots. A total number of 1,625 documents published from 1999 2018 were screened to conduct quantitative analysis, which were retrieved in the Web of Science database with defined search terms. This paper can enable researchers to gain wider and deeper insight into affective computing researches in the last decades through bibliometric analysis, thereby facilitating relevant researchers having a general understanding of aggregate performance in the affective computing field and finding research directions in the future.