OBJECTIVE Retrograde cerebral perfusion (RCP) is a brain protection technique that is adopted generally for anticipated short periods of deep hypothermic circulatory arrest (DHCA). However, the real impact of this technique on cerebral protection during DHCA remains a controversial issue. METHODS For 344 (59.5%) of 578 consecutive patients (mean age, 66.9 ± 10.9 years) who underwent cardiovascular surgery under DHCA at the present authors' institution (1999-2015), RCP was the sole technique of cerebral protection that was adopted in addition to deep hypothermia. Surgery of the thoracic aorta was performed in 95.9% of these RCP patients; in 92 cases there was an aortic arch involvement. Outcomes were reviewed retrospectively. The focus was on postoperative neurological dysfunctions. RESULTS There were 33 (9.6%) in-hospital deaths. Thirty-one (9%) patients had permanent neurological dysfunctions and 66 (19.1%) transitory neurological dysfunctions alone. Age older than 74 years (odds ratio [OR], 1.88, P = .023), surgery for acute aortic dissection (OR, 2.57; P = .0009), and DHCA time longer than 25 minutes (OR, 2.44; P = .0021) were predictors of neurological dysfunctions. The 10-year nonparametric estimate of freedom from all-cause death was 61.8% (95% confidence interval, 57.8%-65.8%). Permanent postoperative neurological dysfunctions were risk factors for cardiac or cerebrovascular death (hazard ratio, 2.6; P = .039) even after an adjusted survival analysis (P < .04). CONCLUSIONS According to the study findings, RCP, in addition to deep hypothermia, combines with a low risk of neurological dysfunctions provided that DHCA length is 25 minutes or less. Permanent postoperative neurological dysfunctions are predictors of poor late survival.