PSA scores, a general overview

2/11/20264 min read

New Jersey’s Public Safety Assessment (PSA) scoring system is a data‑driven, algorithmic tool used to inform pretrial release and detention decisions under the state’s Criminal Justice Reform Act (CJRA), which took effect on January 1, 2017. The PSA replaces cash‑based bail with a risk‑based framework designed to reduce unnecessary pretrial incarceration while maintaining public safety and court‑appearance rates. For lawyers and non-lawyers alike, understanding how the PSA assigns scores, what those scores mean, and how they interact with the New Jersey‑specific Decision‑Making Framework (DMF) is essential to navigating the pretrial process.

Purpose and Legal Framework

The PSA is an actuarial instrument that estimates the likelihood of three outcomes if a defendant is released prior to trial: failure to appear (FTA), new criminal activity (NCA), and new violent criminal activity (NVCA). New Jersey adopted the PSA as part of statewide reforms that eliminated the traditional cash‑bail model and instead required judges to determine whether a defendant could be released on conditions or should be detained solely by reason of risk. The PSA is not a standalone decision‑maker; it informs a judge’s assessment under New Jersey’s DMF, which translates PSA scores into recommended levels of supervision or detention.

Under the CJRA, a PSA is generated shortly after arrest, and its results are used at multiple points: at the initial arrest decision, during the first court appearance, and at any detention or modification hearing. The system presumes that accused individuals have a right to liberty while preparing their defense, but that liberty may be restricted if the defendant’s PSA‑informed risk profile justifies supervision or detention.

How the PSA Generates Scores

The PSA relies exclusively on objective, criminal‑history‑based information from state databases and explicitly excludes race, ethnicity, gender, socioeconomic status, and other demographic characteristics. The system uses nine risk factors, each weighted based on its statistical relationship with FTA and NCA. These factors include age at the time of arrest, prior failures to appear, prior convictions, prior arrests without convictions, pending charges, and the nature of the current offense, among others.

For each defendant, the PSA computes raw scores for FTA and NCA, then converts them to scales of 1–6, with 6 representing the highest risk. A separate “flag” indicates whether the algorithm detects an elevated risk of a new violent crime (NVCA), which can trigger more stringent DMF recommendations even if the numeric FTA/NCA scores are moderate. Because the PSA is algorithmic, small differences in the underlying data—such as disputed prior‑record entries or incorrectly attributed arrests—can materially affect the final score, creating a practical avenue for defense counsel to challenge the accuracy of the inputs rather than the algorithm itself.

Scoring Bands and Pretrial Monitoring Levels

New Jersey’s DMF groups PSA scores into broad risk bands and maps them to recommended pretrial monitoring levels, which in turn guide conditions of release. Although the precise mapping can vary slightly by county and over time, the basic structure is widely consistent. For example, low FTA/NCA scores (often around 1–2) typically correspond to Release on Recognizance (ROR) or minimal conditions, while higher scores (4–6) may justify graduated supervision levels or detention.

Most jurisdictions describe three primary monitoring levels. Level I is the least restrictive, often involving monthly telephone check‑ins with Pretrial Services and basic address‑verification requirements. Level II may require bi‑weekly or monthly in‑person reporting, travel restrictions, and sometimes a curfew. Level III and Level III+ involve more intensive supervision, including weekly reporting, substance‑abuse testing, mental‑health treatment referrals, and, in some cases, electronic monitoring via GPS tether, usually reserved for higher‑risk cases.

Role of the Judge and Prosecutor

Although the PSA and DMF provide structured recommendations, New Jersey law preserves judicial discretion. At the first appearance and at any subsequent detention hearing, the judge considers the PSA scores alongside the nature of the current charges, the defendant’s ties to the community, criminal‑history nuances, and any interventions or mitigating factors presented by counsel. Prosecutors may request pretrial detention if the PSA indicates a substantial risk of FTA, NCA, or NVCA, and the judge must weigh that request against the presumption of release embedded in the CJRA.

Research on New Jersey’s reforms indicates that the PSA has contributed to a marked reduction in pretrial jail populations without a corresponding increase in reoffending or failure‑to‑appear rates. However, critics—particularly in the criminal‑defense and civil‑liberties communities—have raised concerns about algorithmic transparency, potential proxy discrimination, and the risk that high scores could function as a de facto “risk‑based bail” system, even though cash bail is largely eliminated.

Practical Implications for Counsel

For attorneys, the PSA creates several strategic imperatives. First, early case review must include a careful audit of the underlying data used to generate the score, such as prior arrests, convictions, and pending charges, to identify and move to correct any inaccuracies. Second, counsel should be prepared to present contextual information—employment, family ties, housing stability, and treatment history—that can shape the judge’s interpretation of the score even when the DMF points toward a more restrictive recommendation.

Third, because the NVCA flag can trigger heightened scrutiny, it is critical to explore whether the arrest or charge history truly supports a violence‑risk profile, and to emphasize mitigating factors if the defendant’s actual conduct is non‑violent or situational. Finally, defense counsel should be familiar with the local DMF chart and county‑specific practices, since the same PSA score may translate into different levels of supervision or no‑contact conditions depending on jurisdiction and prosecutorial policy.

In sum, New Jersey’s Public Safety Assessment scoring system represents a major shift from a monetary‑bail model to a risk‑informed, data‑driven pretrial framework. By assigning numeric scores for failure‑to‑appear and new‑criminal‑activity risk, and flagging elevated violence risk, the PSA helps judges balance liberty interests with public‑safety concerns. For criminal‑defense attorneys, understanding how PSA scores are generated, how they interact with the DMF, and how to challenge or contextualize them is essential to securing the least-restrictive pretrial outcome for their clients.