Mass incarceration is increasingly recognized as a driver of poor health and health inequities, yet little is known about the health effects of decarceration, the process of reducing incarceration rates. Study of decarceration’s health effects is hindered by the difficulty of systematically identifying where, when, and how decarceration has occurred. Policy surveillance is difficult to scale; legislative reforms do not always translate into reduced incarceration; and de facto changes (e.g. prosecutorial discretion, fiscal constraints) are often overlooked. To address this gap, we developed a scalable approach to identify county-level decarceration using publicly available administrative data. Applying joinpoint regression to longitudinal jail and prison measures from over 2,870 U.S. counties (1999-2019), we defined four operational types of decarceration: reduced pretrial detention, reduced jail time, reduced prison admissions, and reduced prison time. Seventy percent of counties experienced at least one decarceration type during the study period. Reduced prison admissions and reduced pretrial detention were most common, each occurring in about 40% of counties. Frequency and timing varied markedly by region, state, and urbanicity. Declines were typically modest (median ten-year reductions of 19-35%) and often followed recent growth to above-average incarceration levels, suggesting reactions to unsustainable growth rather than proactive structural reforms. Validation against documented instances of real-world decarceration indicated alignment with known drivers while demonstrating the approach’s ability to detect within-state heterogeneity following state-level reforms. This study provides the first systematic characterization of county-level decarceration in the U.S. over two decades and provides a framework and hypothesis-generating resource to support comparative and quasi-experimental studies of decarceration’s heterogeneous health effects.