The Self-Administered Interview (SAI©) is designed to elicit detailed witness reports in the aftermath of incidents. In two sets of meta-analyses, we compared the number of correct details reported, the number of incorrect details reported, and the accuracy of reports provided by witnesses in initial reports (SAI© vs. other reporting formats) and in subsequent accounts (initial SAI© vs. no initial SAI©). The number of comparisons ranged from 15 to 19, (N = 722 to 977). For initial accounts, the SAI© was associated with more correct details and more incorrect details than other reporting formats; accuracy was slightly lower for the SAI© than for other reporting formats. Subsequent accounts were more detailed and accurate for witnesses who had completed an initial SAI© than for those who had not. The SAI© is an effective tool for capturing detailed initial accounts and for preserving witness memory until a formal interview can be conducted.