Risk assessments and classification tiers shape where a person can live, how they are supervised, whether they appear online, and how long they must remain on the registry. Yet despite their enormous influence, most people — including those forced to live under them — don’t understand what these labels actually mean.
This explainer breaks down how risk levels are created, why they differ so dramatically between states, and what they really mean for day-to-day life.
When lawmakers created classification systems in the 1990s and 2000s, they claimed they were building a scientific way to identify people who might pose ongoing danger. But most current systems are not scientific at all — and some aren’t even based on risk.
A. The federal system (SORNA) is offense-based, not risk-based
The federal Sex Offender Registration and Notification Act (SORNA) assigns Tier I, II, or III solely based on the conviction statute, not on personal characteristics, treatment progress, or any actuarial risk tool.
The DOJ SMART Office is explicit: SORNA tiers are “offense-based, not risk-based.” Source: SMART Office Guide to SORNA
A Tier III designation does not mean someone is a “high risk” to reoffend — it simply means Congress placed their statute in the Tier III bucket.
Most people assume federal SORNA controls their registration level, their housing, and whether they appear online. It doesn’t.
What SORNA does:
- Sets baseline federal standards for how states should register people
- Defines Tier I, II, III using statutes — not risk
- Specifies minimum durations (15 / 25 / lifetime)
- Requires jurisdictions to share information
- Applies to states, tribes, territories, and federal agencies
What SORNA does NOT do:
- SORNA does not force states to adopt its tiers. Source: CRS Report R43954
- SORNA does not override state law.
- SORNA does not determine public posting.
- SORNA does not impose residency restrictions.
- SORNA does not control probation or parole.
- SORNA tiers don’t automatically follow you across states.
- SORNA governs jurisdictions, not individuals.
Bottom line: Your day-to-day experience is shaped by state law, not SORNA — and the two systems often contradict each other.
“Your risk level often reflects your ZIP code more than your actual risk.”
A. How long you must register
SORNA sets federal minimums:
- Tier I: 15 years
- Tier II: 25 years
- Tier III: Lifetime
Source: SORNA Final Rule (Federal Register, 2021)
But states routinely create longer durations or systems that ignore SORNA entirely.
B. Whether you appear on the public website
Some states list everyone; others list only certain levels. There is no national standard, and being “high risk” doesn’t always mean public posting.
Source: NIJ: Evaluating Effectiveness
C. Housing restrictions and residency zones
Residency bans are among the most damaging impacts of classification systems. CASOMB found they increased homelessness and risk factors, without reducing reoffending.
Sources: CASOMB: Residence Restrictions (2011) and CASOMB: Homelessness Report (2008)
“A risk label doesn’t just follow you — it rearranges your whole life.”
A. Group predictions ≠ individual predictions
Static-99R was designed to predict how groups behave — not individuals.
Source: Static-99R Coding Rules
B. One point can change your life
A tiny numerical difference in score can shift someone into a “high-risk” category — even when the real-world risk difference is negligible.
C. Risk drops sharply over time
Desistance research shows that after 10–15 years offense-free, reoffense risk declines dramatically and becomes similar to nonsexual felonies.
Source: The Sentencing Project (2024)
D. Some tools overpredict the “high-risk” category
Minnesota DOC found MnSOST-3.1 overpredicted reoffense in its highest category.
Source: MnSOST-3.1 Technical Report
“Risk isn’t a lifetime identity — it changes, and the science proves it.”
- “I was Level 1, now I’m Level 3.”
- “Why did my duration increase?”
- “Why am I online here but not there?”
The answer: state and federal systems operate independently. Source: CRS R43954
“Cross a state line and your ‘risk level’ may change — even though you haven’t.”
A. Homelessness is the most documented outcome
CASOMB found an 800% increase in homelessness among registrants after residency bans were enacted.
B. High-risk labels often create instability, not safety
People labeled “high-risk” often endure community alerts, harassment, severe housing limits, and job discrimination — instability reduces public safety.
C. Families are caught in the blast radius
Partners and children often face landlord rejections, loss of housing, and social harassment.
“Unstable people don’t make safer communities — stable people do.”
- 92.3 % of people released for a sexual offense were not rearrested for another sexual offense.
- People with sexual convictions had lower overall recidivism than most other felony groups.
Source: BJS Recidivism Study (2005–2014)
A 25-year meta-analysis reached similar conclusions. Source: Zgoba & Mitchell Meta-analysis (2021)
“The data is clear: most people with sexual convictions do not reoffend.”
- Your classification may not reflect who you are.
- Your level may change if you move states.
- Downward movement is uncommon.
- Stability matters far more than any actuarial score.
- A full life is still possible.
“A risk score is not your future — your stability is.”
Risk classifications were originally marketed as public-safety tools. But today they often overstate danger, ignore desistance, create homelessness, destabilize families, vary wildly between states, mismatch the science, and function as punishment, not prevention. A safer, saner system would emphasize stability, support, and evidence — not static labels or distorted predictions.
- SMART Office Guide to SORNA
- Congressional Research Service R43954
- SORNA Final Rule (Federal Register)
- NIJ: Evaluating Effectiveness
- CASOMB: Residence Restrictions (2011)
- CASOMB: Homelessness Report (2008)
- Static-99R Coding Rules (Public Safety Canada)
- MnSOST-3.1 Technical Report
- The Sentencing Project (2024)
- BJS Recidivism Study (2005–2014)
- Zgoba & Mitchell Meta-analysis (2021)