Acupressure is a non-invasive treatment in which pressure is applied to specific body points. Following public health concerns about poor sleep quality and increasing interest in alternative treatments, a systematic review and a meta-analysis were designed to evaluate the effects of acupressure on the quality of sleep. Ten English (PubMed, CENTRAL, CINAHL etc.) and five Chinese (CNKI, WANFANG etc.) databases were searched and the validity of the eligible studies was critically appraised. Thirty-two eligible randomized controlled trials of moderate to high quality which employed polysomnography, actigraphy, or self-assessment sleep quality tools were included. We conducted a meta-analysis using a random effects model with the Pittsburgh sleep quality index (PSQI) as the primary outcome measure (968 adult patients, 13 trials) for trials investigating the effects of traditional Chinese medicine acupressure compared with standard and sham treatments. We performed subgroup analyses to detect sources of heterogeneity, identify the use of acupoints in different populations and explore the contributions of PSQI domains to the total score change. Comparison with the sham group (7 trials with 385 patients) yielded low heterogeneity and an overall effect of 13%–19% improvement in the PSQI score (MD = −3.41, 95% CI −4.08, −2.75; I² = 12%). Based on data from four trials (n = 250), sleep latency and sleep duration were most affected. No adverse effects were reported in any of the reviewed trials. Within the limitations of clinical heterogeneity, the results showed that even fragile populations such as the elderly and dialysis patients can benefit from acupressure. Standardized treatment protocols involved 3–5 kg of pressure for one to five minutes per acupoint, delivered three to seven times a week for three to four weeks with the HT7 (Shenmen) acupoint used in most procedures. A high risk of bias due to absence of blinding of patients and personnel remains a serious methodological challenge for acupressure trials and poses a main limitation to presented results. This review has been registered in PROSPERO (registration number CRD42015025013).