MARSSIM Final Status Surveys Explained

A MARSSIM final status survey is the disciplined, statistically defensible way a facility proves that a decommissioned site is clean enough to release, by comparing measured residual radioactivity against derived guideline levels using nonparametric hypothesis tests. It is the evidentiary backbone of license termination.¹³

When a licensee finishes cleaning up a site, the regulator does not accept "it looks clean." It requires a survey designed so that the probability of wrongly releasing a contaminated site is controlled and known. This guide walks through the NUREG-1575 Revision 2 framework — survey-unit classification, DCGLs, the Sign and Wilcoxon Rank Sum tests, and the Data Quality Objectives process — and the worked statistics that decide whether a survey unit passes. DRPS supports decommissioning and license termination through radioactive material license support and radiation safety officer consulting across Florida, Maryland, Virginia, Washington DC, California, and Nevada.¹⁴

Introduction

The purpose of a final status survey is to make a release decision defensible under uncertainty. No survey can measure every square centimeter of a site, so the decision to release always rests on a sample. MARSSIM's contribution is a framework that controls the two ways that sample-based decision can go wrong: releasing a site that is actually contaminated, and failing to release a site that is actually clean.

MARSSIM — the Multi-Agency Radiation Survey and Site Investigation Manual, NUREG-1575, now in Revision 2 (May 2020) — is a consensus document developed jointly by the NRC, EPA, DOE, and DOD.¹ It provides guidance for planning, conducting, evaluating, and documenting building-surface and surface-soil radiological surveys used to demonstrate compliance with dose- or risk-based release criteria.¹

The framework rests on a chain of translation:

A dose limit (25 mrem/yr for unrestricted release under 10 CFR 20.1402) is the regulatory target.²
A DCGL converts that dose limit into a measurable concentration.
Survey-unit classification determines how intensively each area is surveyed.
Nonparametric statistical tests decide whether each survey unit meets the DCGL.
Documentation ties it all together into a license-termination record.

For the surrounding regulatory process, see our companion guide to decommissioning and license termination.

Topic Explanation

From dose limit to DCGL

A regulator's release criterion is a dose: 25 mrem per year to an average member of the critical group.² But a survey meter measures activity concentration, not dose. The bridge between them is the derived concentration guideline level (DCGL) — the residual radioactivity concentration that, through exposure pathway modeling, would produce exactly the dose limit.

MARSSIM distinguishes two DCGLs:

DCGL-W ("W" for wide-area) is the average concentration over a survey unit that corresponds to the dose limit. It is the value used with the statistical tests, which evaluate the survey unit as a whole.
DCGL-EMC (elevated measurement comparison) is applied to small areas. Because a hot spot smaller than the modeled area can exceed the average DCGL without exceeding the dose limit, the DCGL-EMC is a higher, area-adjusted limit used to evaluate individual elevated measurements found during scanning.

Survey-unit classification

MARSSIM does not survey every area the same way. It first classifies areas by their potential for residual contamination, and that classification sets the survey intensity:

Class	Contamination potential	Surface scan coverage	Measurement approach
Class 1	Highest — may exceed the DCGL	100 percent	Systematic measurements on a grid plus full scanning
Class 2	Some potential, not expected to exceed the DCGL	Partial (typically 10–100 percent)	Systematic measurements, less intensive scanning
Class 3	Little or no potential	Judgmental	Judgmental or limited random measurements

Classifying an area too low is the costliest error, because a Class 1 area surveyed as Class 3 can leave contamination undetected. When in doubt, MARSSIM directs the planner to assign the higher class. Survey units are then bounded in size by class (Class 1 units are kept smaller so a fixed number of measurements gives adequate coverage).

The gray region and the null hypothesis

MARSSIM frames the release decision as a hypothesis test. The default null hypothesis is conservative: the survey unit is assumed to exceed the release criterion until the data demonstrate otherwise. This places the burden of proof on showing the site is clean, which is the protective default.

Between the DCGL-W and a lower value called the lower bound of the gray region (LBGR) lies the "gray region," where the consequences of a wrong decision are considered acceptable and where the test is not required to have high power. The width of the gray region, expressed as a relative shift, drives how many measurements are needed.

Key Technical Principles

The relative shift and the number of measurements

The Data Quality Objectives process fixes two error probabilities: the Type I error $α$ (probability of releasing a unit that actually exceeds the criterion) and the Type II error $β$ (probability of failing a unit that is actually acceptable). It also fixes the measurement variability $σ$ . From these, MARSSIM computes the relative shift:

\frac{Δ}{σ} = \frac{DCGL _{W} - LBGR}{σ}

A larger relative shift — a wider gray region relative to the data scatter — makes the two conditions easier to distinguish and requires fewer measurements. A small relative shift requires many measurements to achieve the same confidence.

For the Sign test, the number of measurements per survey unit is:

N = \frac{( Z _{1 - α} + Z _{1 - β} ) ^{2}}{4 ( Sign p - 0.5 ) ^{2}}

where $Z_{1 - α}$ and $Z_{1 - β}$ are standard normal quantiles for the chosen error rates and $Sign p$ is the estimated probability that a random measurement falls below the DCGL-W, obtained from the relative shift. As a worked example, for $α = β = 0.05$ , $Z_{0.95} = 1.645$ , so the numerator is $(1.645 + 1.645)^{2} = 10.8$ ; if the relative shift gives $Sign p = 0.90$ , then:

N = \frac{10.8}{4 ( 0.90 - 0.5 ) ^{2}} = \frac{10.8}{0.64} \approx 17

MARSSIM then applies an adjustment (commonly increasing $N$ by about 20 percent) to protect against missing or unusable measurements, so roughly 20 measurements would be placed in that survey unit. The manual provides lookup tables so planners do not compute this by hand, but the underlying logic is exactly this: error rates and relative shift determine the sample size.¹⁵

Choosing the statistical test

MARSSIM selects between two nonparametric tests based on one question: is the radionuclide of concern present in background?

Situation	Test	Design
Radionuclide NOT in background (e.g., a pure man-made emitter measured specifically)	One-sample Sign test	Compare survey-unit measurements directly against the DCGL-W
Radionuclide present in background (e.g., natural uranium, thorium, or gross activity)	Two-sample Wilcoxon Rank Sum (WRS) test	Compare the survey unit against a matched reference area

Nonparametric tests are used because residual-activity data rarely follow a normal distribution and often include values near or below detection limits; the Sign and WRS tests make no distributional assumption and tolerate a modest number of non-detects.¹⁵

Scanning and the scan MDC

Systematic measurements find the average; scanning finds localized elevated activity between them. A scan is only useful if its minimum detectable concentration (scan MDC) is low enough to detect activity of concern — ideally at or below the DCGL-EMC. The scan MDC depends on the instrument, the scan speed, the background, and the surveyor's detection ability. If the scan MDC cannot get below the DCGL-EMC, the survey design must compensate with a denser measurement grid. Instrument sensitivity and MDC are covered in our guides to minimum detectable activity in contamination surveys and choosing the right radiation survey meter.

Clinical Impact

For a medical, nuclear-pharmacy, or research licensee, MARSSIM is the difference between a smooth license termination and a stalled, costly one. The framework matters most at two moments: when a facility ends a radioactive-material program, and when a regulator reviews the release record.

A hospital nuclear medicine department, a PET radiopharmacy, or a research laboratory that decommissions a hot lab must demonstrate that residual contamination — from spills, decayed sources, or activation — is below the release criterion. Done well, a MARSSIM survey produces a record that a regulator can accept largely as submitted. Done poorly — areas misclassified, DCGLs undocumented, too few measurements, scan MDC above the DCGL-EMC — the survey invites rejection, resurvey, and delay in terminating the license and recovering the facility for other use.

For clearing individual materials and equipment rather than land and building surfaces — a decayed sealed source housing, a contaminated hood, or scrap metal — the companion MARSAME manual (NUREG-1575 Supplement 1) applies the same classification and statistical framework to items instead of areas.⁸

MARSSIM is scalable. Many medical decommissionings qualify for screening DCGLs and involve only a few survey units, so the full statistical machinery is light. But even a simple survey benefits from the MARSSIM discipline: classify the areas, state the DCGLs and their basis, justify the number of measurements, confirm the scan MDC, and document everything. That discipline is what makes the release defensible. For the license-side steps, see radioactive material license renewal and surface contamination limits and equipment release.

Practical Optimization Tips

1. Classify conservatively

When the contamination potential of an area is uncertain, assign the higher class. Under-classification is the error most likely to leave contamination undetected and the hardest to defend later.

2. Document the DCGL basis

State each DCGL, whether it is a screening value or a site-specific dose-modeled value, and the pathway assumptions behind it. A regulator cannot accept a comparison against an undocumented limit.

3. Fix the DQO inputs before measuring

Decide $α$ , $β$ , the DCGL-W, the LBGR, and the expected $σ$ during planning, not after data collection. The number of measurements follows from those choices; setting them afterward invites bias.

4. Verify the scan MDC against the DCGL-EMC

Confirm, before the survey, that the chosen instrument and scan speed can detect activity at or below the DCGL-EMC. If not, add measurements or change the instrument.

5. Keep reference areas comparable

For WRS surveys, the reference area must match the survey unit in the background radionuclide, media, and physical characteristics. A poorly matched reference area invalidates the comparison.

Common pitfalls to avoid

Under-classifying survey units to reduce survey effort — the single most consequential design error.
Using a screening DCGL where site conditions violate its assumptions, making the limit inapplicable.
Collecting too few measurements because the DQO inputs were never fixed.
Scanning with an MDC above the DCGL-EMC, so hot spots go undetected.
Poor documentation — an otherwise valid survey that cannot be reconstructed from its record is not defensible.

Regulatory Considerations

A MARSSIM final status survey demonstrates compliance with the radiological criteria for license termination in Subpart E of 10 CFR Part 20. The survey is the evidence; the regulation is the standard it must meet.

10 CFR 20.1402 sets the unrestricted-release criterion: 25 mrem/yr TEDE to an average member of the critical group plus ALARA.²
10 CFR 20.1403 provides the alternative criteria for restricted release, used when unrestricted release is not achievable, with additional dose limits and institutional controls.⁶
NUREG-1575, Revision 2 (MARSSIM) is the accepted methodology for designing and evaluating the survey, and consensus clearance standards such as ANSI/HPS N13.12 provide complementary screening levels for surface and volume radioactivity.¹⁹
NUREG-1757, the Consolidated Decommissioning Guidance, ties the survey into the broader NRC decommissioning and license-termination review, including dose modeling and screening DCGLs.⁴
NUREG-1505 provides the nonparametric statistical methodology behind the Sign and WRS tests, and NUREG-1507 provides minimum detectable concentrations for typical survey instruments.⁵⁷

Agreement States administer equivalent programs. Of the states DRPS serves, Florida, Maryland, Virginia, California, Nevada, Pennsylvania, New York, and New Jersey are NRC Agreement States that regulate license termination under their own radiation-control rules, while Washington, DC and Delaware are regulated directly by the NRC for radioactive material. A licensee must confirm which authority reviews its decommissioning and which DCGLs and documentation it will accept before the survey is designed. DRPS integrates final status survey planning into medical physics consulting and radioactive material license support.

Frequently Asked Questions (FAQs)

What is a MARSSIM final status survey?

A final status survey (FSS) is the radiological survey that demonstrates a site meets the release criterion after decommissioning. MARSSIM, the Multi-Agency Radiation Survey and Site Investigation Manual (NUREG-1575, Revision 2), provides the standardized method for planning, conducting, evaluating, and documenting that survey so a regulator can accept the release of the site or building for unrestricted use.

What release limit does MARSSIM demonstrate compliance with?

For NRC and Agreement State licensees, unrestricted release under 10 CFR 20.1402 requires that residual radioactivity distinguishable from background produce a total effective dose equivalent no greater than 25 mrem (0.25 mSv) per year to an average member of the critical group, and that the residual radioactivity be reduced to levels that are as low as reasonably achievable. MARSSIM translates that dose limit into measurable concentration guideline levels.

What is a DCGL?

A derived concentration guideline level, or DCGL, is the residual radioactivity concentration that corresponds to the release dose limit. The DCGL-W is the wide-area average used with the statistical tests over a survey unit; the DCGL-EMC is the elevated measurement comparison value applied to small areas or hot spots. DCGLs convert an abstract dose limit into a number that can be measured in the field.

How are survey units classified in MARSSIM?

MARSSIM classifies areas by their potential for residual contamination. Class 1 areas have the highest potential to exceed the DCGL and receive the most intensive survey, including 100 percent surface scanning. Class 2 areas are not expected to exceed the DCGL but have some potential and receive partial scanning. Class 3 areas have little or no potential and receive judgmental measurements. Classification drives how many measurements and how much scanning a survey unit needs.

What statistical tests does MARSSIM use?

MARSSIM uses nonparametric hypothesis tests. When the radionuclide of concern is not present in background, the one-sample Sign test is used. When the radionuclide is present in background, the two-sample Wilcoxon Rank Sum test compares the survey unit against a reference area. Both tests decide whether the survey unit meets the release criterion at a chosen confidence level.

What is the Data Quality Objectives process?

The Data Quality Objectives (DQO) process is a seven-step planning method that defines the decision to be made, the acceptable error rates, and the data needed to make it. In MARSSIM it fixes the Type I and Type II error probabilities, the gray region between the release limit and the lower bound of the gray region, and from those the number of measurements a survey unit requires.

Does MARSSIM apply to a small medical facility?

MARSSIM is scalable. A small medical or nuclear-pharmacy decommissioning may use screening DCGLs and a modest number of survey units, while a large site needs the full statistical design. Even when the survey is simple, following the MARSSIM framework for classification, DCGLs, and documentation makes the license-termination record defensible to the NRC or Agreement State.

Key Takeaways

A final status survey proves a decommissioned site meets its release criterion using a statistically defensible sample, not a visual impression.
DCGLs translate the 25 mrem/yr dose limit into measurable concentrations — DCGL-W for the survey-unit average, DCGL-EMC for localized elevated measurements.
Survey-unit classification (Class 1, 2, 3) sets survey intensity, and under-classification is the most consequential design error.
Nonparametric tests decide the release: the Sign test when the radionuclide is absent from background, the WRS test when it is present.
The Data Quality Objectives process fixes the error rates and the relative shift, which together determine the number of measurements per survey unit.
Scanning must have an MDC at or below the DCGL-EMC, or the design must add measurements to compensate.

Conclusion

MARSSIM turns "prove it is clean" into a defined, repeatable procedure. By translating a dose limit into DCGLs, classifying areas by contamination potential, fixing acceptable error rates through the Data Quality Objectives process, and deciding each survey unit with a nonparametric test, it produces a release decision whose reliability is quantified rather than asserted.

For a medical or research licensee, the payoff is a license-termination record that survives regulatory review the first time. The framework scales from a single decommissioned hot lab to a full site, but the discipline is the same at every scale: classify honestly, document the DCGLs, justify the measurements, confirm the scan sensitivity, and record it all. That discipline is what lets a facility close a radioactive-material program cleanly and move on.

How DRPS Can Help

Diagnostic Radiation Physics Services supports decommissioning and license termination end to end: survey-unit classification, DCGL selection and documentation, Data Quality Objectives planning, instrument and scan-MDC verification, final status survey design and data assessment, and the release record a regulator will accept. This work is delivered through radioactive material license support, radiation safety officer consulting, and medical physics consulting.

DRPS supports facilities across our service locations, including Florida, Maryland, Virginia, Washington DC, California, Nevada, New York, Pennsylvania, New Jersey, and Delaware.

Related Resources

References

U.S. Nuclear Regulatory Commission, EPA, DOE, DOD. Multi-Agency Radiation Survey and Site Investigation Manual (MARSSIM), NUREG-1575, Revision 2. 2020. nrc.gov
U.S. Nuclear Regulatory Commission. 10 CFR 20.1402: Radiological criteria for unrestricted use. nrc.gov
U.S. Environmental Protection Agency. Multi-Agency Radiation Survey and Site Investigation Manual (MARSSIM). epa.gov
U.S. Nuclear Regulatory Commission. NUREG-1757: Consolidated Decommissioning Guidance. nrc.gov
U.S. Nuclear Regulatory Commission. NUREG-1505: A Nonparametric Statistical Methodology for the Design and Analysis of Final Status Decommissioning Surveys. nrc.gov
U.S. Nuclear Regulatory Commission. 10 CFR 20.1403: Criteria for license termination under restricted conditions. nrc.gov
U.S. Nuclear Regulatory Commission. NUREG-1507: Minimum Detectable Concentrations with Typical Radiation Survey Instruments for Various Contaminants and Field Conditions. nrc.gov
U.S. Nuclear Regulatory Commission. Multi-Agency Radiation Survey and Assessment of Materials and Equipment Manual (MARSAME), NUREG-1575 Supplement 1. nrc.gov
Health Physics Society. ANSI/HPS N13.12: Surface and Volume Radioactivity Standards for Clearance. hps.org