Background: Secondary cerebrospinal leaks (CSF) are leaks that recur after an initial endoscopic repair of CSF leaks. Identification of characteristics that could predict secondary leaks may allow surgeons to plan repairs with the knowledge that these defects are more likely to fail. Objective: To identify characteristics that could predict secondary CSF leaks. Methods: A search of all studies that reported outcomes after endoscopic repair of CSF leaks was conducted by using medical literature data bases. Studies with the following criteria were included: written in the English language, secondary CSF leaks after primary repair, and assessment of success of fistula repair. Data extracted included the etiology of the leak, site of the leak, reconstructive materials used, and success of the repair. Results: Ninety-four studies, from 1988 to 2015, with a total of 3149 primary CSF leaks were included. Sixty-three studies (67%) had a success rate of ≥80%; 77 studies (82%) had secondary leak rates of 40%. For 88 of the 94 studies (94%), the success rate after secondary repair improved to 81.0-100%. Of 48 studies that included leaks of more than one site, the sphenoid was the most common site of leak in 26 studies (55%), the ethmoid was the most common site in 16 studies (34%), and the cribriform was the most common site in 11 studies (23%). Conclusion: The rate of secondary CSF leaks was low due to advances in endoscopic repair techniques. Spontaneous and iatrogenic CSF leaks were more likely to recur, especially without adequate control of underlying factors, such as increased intracranial pressure and obesity. Further studies with consistent reporting are required for more definitive conclusions about secondary CSF leaks.