Objective To investigate the effects of unilateral thoracic paravertebral block (TPVB) on anesthesia and postoperative analgesia in thoracoscopic mediastinal surgery. Methods This study was a single‑center retrospective cohort study. A total of 522 patients underwent mediastinal or thymic surgery from 2016 to 2018 in Peking University People's Hospital. According to inclusion criteria, 200 patients were enrolled in the study, including 119 patients receiving paravertebral nerve block (group P) and 81 patients not receiving paravertebral nerve block (group S). The following data were collected: the doses of sufentanil, remifentanil and other anesthetics during operation, the use of ephedrine and norepinephrine, the type of operation, intraoperative blood loss, urinary volume, and infusion volume. The following postoperative data were recorded: the dosage of sufentanil 1 d and 3 d after operation, the Numerical Rating Scale (NRS) score at rest and during movement 1 and 3 days after operation, the number of analgesic pump compression, the incidences of nausea and vomiting, and the incidence of dizziness. Results Compared with group S, group P required less intraoperative doses of sufentanil and remifentanil than group S (P<0.05) and presented more urinary volume and infusion volume (P<0.05), with an increased percentage of patients to use pressure‑raising agents during operation (P<0.05). Furthermore, compared with group S, group P presented decreased length of postanesthesia care unit (PACU) stay (P<0.05), and required a reduced number of analgesic pump compression on day 1 after surgery (P<0.05) and decreased dosage of sufentanil on day 1 after operation (P<0.05), with decreases in the incidence of nausea and vomiting on day 1 after operation (P<0.05) and rescue analgesia (P<0.05). There were no statistical differences between the two groups in intraoperative blood loss, the length of operation, the incidences of preoperative hypertension, hypotension, arrhythmia, medium to severe pain, nausea and vomiting, NRS at rest and during movement days 1, 2 and 3 after surgery, and the dosages of sufentanil days 2 and 3 after surgery (P>0.05). Conclusions The application of paravertebral nerve block before thoracoscopic mediastinal surgery can reduce the doses of opioids used during operation, shorten the length of PACU stay, reduce the doses of opioids used after operation, and reduce the incidence of nausea and vomiting on day 1 after operation, but may increase the incidence of intraoperative hypotension.