["Criminology &Public Policy, Volume 25, Issue 2, Page 213-240, May 2026. ", "\nAbstract\n\nResearch summary\nIndiana Code allows local prosecutors to withhold prosecution and offer diversion for residents charged with low‐level offenses. In doing so, prosecutors can minimize the collateral consequences of justice‐system involvement. Although widely adopted as an alternative to traditional prosecution, few studies have evaluated the effectiveness of prosecutor‐led diversion programs. The current study builds on this area of research by evaluating two prosecutor‐led diversion programs operating in one large and one moderately sized jurisdiction, highlighting their effectiveness in limiting future criminal justice contact. We examined administrative prosecution records from 2013 to 2023 in which 21,653 filed cases met statutory eligibility for diversion, and 4572 individuals accepted a diversion offer in the larger site. In the moderate‐sized jurisdiction, 17,357 cases were eligible, with 6335 offers accepted. Propensity score‐weighted and instrumental variable regression techniques indicate that diversion was associated with a reduced probability of a new criminal case filing 1‐year after the instant case filing relative to all other eligible cases. Moreover, program participation moderated the effect of race on future criminal justice contact.\n\n\nPolicy implications\nProsecutor‐led diversion programs are one strategy used to efficiently manage criminal caseloads, hold defendants accountable, temper the collateral consequences of conviction, and reduce recidivism. The findings reported here indicate that these programs are associated with lower rates of recidivism. However, the programs’ discretionary design results in case‐ and defendant‐level sorting and selection processes that shape program efficacy. After adjusting for sorting and selection effects, associations between diversion and favorable outcomes persist but raise important questions about both equitable access to diversion programs and the reliability of previously reported diversion findings.\n\n"]