Objective The aim of this study was to examine prospective memory (PM) and its socio-demographic, clinical, and neurocognitive correlates in first episode schizophrenia (FES). Methods Fifty-one FES patients and 42 healthy controls formed the study sample. Time- and event-based PM (TBPM and EBPM) performance were measured with the Chinese version of the Cambridge Prospective Memory Test (C-CAMPROMPT). A battery of neuropsychological tests was also administered. Patients' clinical symptoms were evaluated with the Positive and Negative Symptom Scale (PANSS). Results Patients performed significantly worse in both TBPM (8.7 ᠵ.3 vs. 14.8 ᠳ.5) and EBPM (11.3 ᠴ.7 vs. 15.7 ᠲ.7) than the controls. After controlling for age, gender, education level and neurocognitive test score, the difference in performance on the two types of PM tasks between patients and controls was no longer present. In multiple linear regression analyses, longer duration of untreated psychosis (DUP), lower scores of the Hopkins Verbal Learning Test-Revised (HVLT-R) and the categories completed of the Wisconsin Card Sorting Test (WCST-CC) and higher score of the Color Trails Test-2 (CTT-2) contributed to poorer TBPM performance, while lower score of HVLT-R, higher score of the perseverative errors of the Wisconsin Card Sorting Test (WCST-PE) and longer DUP contributed to worse performance on EBPM. Conclusions Both subtypes of PM are impaired in first-episode schizophrenia suggesting that PM deficits are an integral part of the cognitive dysfunction in the disease process.