The DOI matches its Crossref record; 33 of the 34 references it cites were found in a registry or an index; none of the 14 sources its claims cite is checked for support yet.
A statistical test suite for random and pseudorandom number generators for cryptographic applications
Open accessAbstract
The Information Technology Laboratory (ITL) at the National Institute of Standards and Technology (NIST) promotes the U.S. economy and public welfare by providing technical leadership for the nation's measurement and standards infrastructure. ITL develops tests, test methods, reference data, proof of concept implementations, and technical analysis to advance the development and productive use of information technology. ITL's responsibilities include the development of technical, physical, administrative, and management standards and guidelines for the cost-effective security and privacy of sensitive unclassified information in Federal computer systems.
Overview
CoChat Summary
TL;DR
This NIST Special Publication 800-22 Revision 1a presents a suite of 15 statistical tests for evaluating whether binary sequences produced by random or pseudorandom number generators are suitable for cryptographic applications. It provides test purposes, mathematical descriptions, examples, and guidance for interpreting P-values, while cautioning that statistical testing is only a first step and cannot replace cryptanalysis.
Key findings
- The suite consists of 15 tests: Frequency (Monobit), Frequency within a Block, Runs, Longest Run of Ones, Binary Matrix Rank, Discrete Fourier Transform (Spectral), Non-overlapping Template Matching, Overlapping Template Matching, Maurer’s “Universal Statistical,” Linear Complexity, Serial, Approximate Entropy, Cumulative Sums, Random Excursions, and Random Excursions Variant. The Frequency test is recommended first because it supplies the most basic evidence of non-randomness.
- Each test uses the null hypothesis \(H_0\) that the sequence is random; if P-value \(\ge \alpha\), the sequence is accepted as random, and if P-value \(<\alpha\), it is rejected. Examples use \(\alpha=0.01\), and NIST recommends \(\alpha\) in the range [0.001, 0.01].
- The document states that no finite set of statistical tests is “complete,” that statistical testing cannot absolutely certify a generator, and that statistical testing cannot serve as a substitute for cryptanalysis.
- Sample data files are 1,000,000 bits (binary expansions of \(e\), \(\pi\), \(\sqrt{2}\), \(\sqrt{3}\), and a G-SHA-1 sequence); reported example P-values include Frequency for \(\pi\) = 0.578211, for \(e\) = 0.953749, and for G-SHA-1 = 0.604458.
Methods
- Design: The NIST Statistical Test Suite is an ANSI C reference implementation of 15 tests; each test defines a test statistic, a reference distribution (e.g., normal, half-normal, chi-square), a P-value, and a decision rule at the 1% level.
- Data or sample: Tests apply to binary sequences from hardware or software RNGs/PRNGs; sample files are 1,000,000 bits, while many worked examples use small illustrative lengths such as \(n=100\).
- Main analysis: For each sequence, P-values are computed and compared to \(\alpha\); empirical interpretation examines the proportion of sequences passing a test and the uniformity of P-values. A multiple-test dependence study applied 161 test P-values to Blum-Blum-Shub sequences of length 1,000,000 and used principal components/factor analysis, concluding there was no large redundancy among tests.
Limitations
- The paper states that no set of statistical tests can absolutely certify a generator as appropriate, that statistical testing cannot substitute for cryptanalysis, and that generator design and cryptanalysis are outside its scope.
- The paper states that many examples use small sample sizes for illustration only, e.g., \(n=10\), where the normal approximation is not really applicable; many tests rely on asymptotic reference distributions for large \(n\) (order \(10^3\) to \(10^7\)), and input parameter constraints apply.
- The text makes evident additional limitations: numerical precision can degrade for large incomplete gamma parameter \(a\); Maurer’s variance approximation can make the test 2.67 times more permissive than theoretically admitted; Random Excursions is discontinued if \(J<500\); and multiple tests can yield contradictory conclusions, as in the Random Excursions example where one state gave a non-random result while seven states gave random results.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures
Verification
Claims the paper makes
0 Valid · 1 Not valid · 11 Not checked · 13 Own results and interpretations
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The asymptotic distribution of ( Ln -µ n ) / σ n along the sequence of even or odd values of n is that of a discrete random variable obtained via a mixture of two geometric random variables (one of them taking only negative values). Strictly speaking, the asymptotic distribution as such does not exist.”
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, no set of statistical tests can absolutely certify a generator as appropriate for usage in a particular application, i.e., statistical testing cannot serve as a substitute for cryptanalysis.”
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This graph showed that there is no large redundancy among our tests.”
In the paper
“Ueli Maurer of the Department of Computer Science at Princeton University introduced this test in 1992. Maurer's test statistic relates closely to the per-bit entropy of the stream, which its author asserts is 'the correct quality measure for a secret-key source in a cryptographic application.'”
What the sources say
- Not matchedcites Maurer
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Not matched
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“The test rejects the randomness hypothesis immediately if J is too small, i.e., if the following P-value is small:”
20 more claims
Create a free account to see all 25 claims and every source verdict.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“However, the latest version 2.1.2 [1], released in 2014, does not incorporate several modifications.”
From Checking the quality of approximation of p-values in statistical tests for random number generators by using a three-level test (2019) · cites this paper as [1]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“The NIST test suite consists of 15 statistical tests for randomness of bit sequences, and its result is a list of 188 p -values.”
From Checking the quality of approximation of p-values in statistical tests for random number generators by using a three-level test (2019) · cites this paper as [1]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“Note that the current implementation of the NIST test suite uses one-level or two-level tests, differently from the above experiments”
From Checking the quality of approximation of p-values in statistical tests for random number generators by using a three-level test (2019) · cites this paper as [1]
The cited source says what the claim attributes to it.
In the citing paper
“NIST SP-800.22 statistical test suite [90] and Die-hard test suite [91] and Test U01 [92] and AIS31 [93] are most common statistical test suites to evaluate their RNGs by researchers.”
From True Random Number Generators Using Electrical Noise (2019) · cites this paper as [90]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 2.3.8 Example
“(input) τ = 0.02 (processing) π = 0.42 (processing) Vn(obs) = 52 (output) P-value = 0.500798 (conclusion) Since P-value ≥ 0.01 , accept the sequence as random.”
§ 2.3.4 Test Description
“If it can be shown that π -1 ≥τ , then the 2 Runs test need not be performed (i.e., the test should not have been run because of a failure to pass test 1, the Frequency (Monobit) test). If the test is not applicable, then the P-value is set to 0.0000.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 2.7.4 Test Description
“| 8-10 | 010 (hit) | 2 | 110 | 1 |”
§ 2.7.4 Test Description
“For the above example, if m = 3 and the template B = 001, then the examination proceeds as follows:”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 2.9.3 Test Statistic and Reference Distribution
“f n : The sum of the log2 distances between matching Lbit templates, i.e., the sum of the number of digits in the distance between Lbit templates.”
§ 3.9 Maurer's 'Universal Statistical' Test
“Then it averages over all the expansion lengths by the number of test blocks.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Input Parameters:
“The parameters p , q and s are not fixed in code; thus, the user will not be able to reconstruct the original sequence because these values will vary (i.e., they are dependent on the system time).”
§ Input Parameters:
“The input parameters have been fixed so the code will reproduce the same results.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 5.5.2 Global Parameters
“| MAXNUMOFTEMPLATES | 40 | Non-overlapping Templates Test | | NUMOFTESTS | 16 | Max number of tests | | NUMOFGENERATORS | 12 | Max number of PRNGs |”
§ A.2 Configuration Information
“| 2. | #define MAXNUMOFTEMPLATES | 148 | | 3. | #define NUMOFTESTS | 15 | | 4. | #define NUMOFGENERATORS | 10 |”
2 more findings
Create a free account to see every finding of this check.
Citations
Cites
Handbook of Mathematical Functions
McQuarrie, D. A. (1966) · American Journal of Physics · doi:10.1119/1.1972842 · Cited by 40,433
Numerical recipes in C: the art of scientific computing
(1993) · Choice Reviews Online · doi:10.5860/choice.30-5638 · Cited by 17,933
Rayward-Smith, V. J. (1991) · Journal of the Operational Research Society · doi:10.1057/jors.1991.155 · Cited by 16,821
Handbook of applied cryptography
(1997) · Choice Reviews Online · doi:10.5860/choice.34-4512 · Cited by 10,357
Handbook of Mathematical Functions
(2018) · A Century of Excellence in MEASUREMENTS, STANDARDS, and TECHNOLOGY · doi:10.1201/9781351069397-39 · Cited by 9,062
A universal algorithm for sequential data compression
Ziv, J., & Lempel, A. (1977) · IEEE Transactions on Information Theory · doi:10.1109/tit.1977.1055714 · Cited by 5,456
The Art of Computer Programming
Worlton, W. J. (1968) · Nuclear Science and Engineering · doi:10.13182/nse68-a19548 · Cited by 2,251
Handbook of Applied Cryptography
Menezes, A. J., van Oorschot, P. C., & Vanstone, S. A. (2018) · CRC Press eBooks · doi:10.1201/9780429466335 · Cited by 1,770
Numerical Recipes in C-The Art of Scientific Computing.
G., W., Press, W. H., Flannery, B. P., Teukolsky, S. A., & Vetterling, W. T. (1989) · Mathematics of Computation · doi:10.2307/2008673 · Cited by 1,513
The Fourier Transform and its Applications
Cattermole, K. W. (1965) · Electronics and Power · doi:10.1049/ep.1965.0268 · Cited by 1,417
Cited by
Digital image steganography: Survey and analysis of current methods
Cheddad, A., Condell, J., Curran, K., & Mc Kevitt, P. (2010) · Signal Processing · doi:10.1016/j.sigpro.2009.08.010 · Cited by 1,704
“26] has answered the question “how random is the noise?” qualitatively there are various methods which estimate randomness quantitatively (see [101]).”
Significant-Loophole-Free Test of Bell’s Theorem with Entangled Photons
Giustina, M., Versteegh, M. A. M., Wengerowsky, S., Handsteiner, J., Hochrainer, A., Phelan, K., Steinlechner, F., Kofler, J., Larsson, J.-Å., Abellán, C., et al. (2015) · Physical Review Letters · doi:10.1103/physrevlett.115.250401 · Cited by 1,529
“[30–35], for more information about the experimental setup and data analysis.”
BackgroundRandom numbers certified by Bell’s theorem
Pironio, S., Acín, A., Massar, S., de la Giroday, A. B., Matsukevich, D. N., Maunz, P., Olmschenk, S., Hayes, D., Luo, L., Manning, T. A., et al. (2010) · Nature · doi:10.1038/nature09008 · Cited by 1,420
Handbook of Monte Carlo Methods
Kroese, D. P., Taimre, T., & Botev, Z. I. (2011) · Wiley Series in Probability and Statistics · doi:10.1002/9781118014967 · Cited by 1,104
“2], and adapted version of the Diehard suite of tests by Marsaglia [28], the ones implemented by the National Institute of Standards and Technology (NIST) [37], and various other tests.”
MethodFast physical random bit generation with chaotic semiconductor lasers
Uchida, A., Amano, K., Inoue, M., Hirano, K., Naito, S., Someya, H., Oowada, I., Kurashige, T., Shiki, M., Yoshimori, S., et al. (2008) · Nature Photonics · doi:10.1038/nphoton.2008.227 · Cited by 1,024
“Efforts are being made to develop faster and more reliable generators and establish better standards for random number generation based on stringent tests of randomnes”
Background
Metadata
- Authors
- Bassham, L. E., Rukhin, A. L., Soto, J., Nechvatal, J. R., Smid, M. E., Barker, E. B., Leigh, S. D., Levenson, M., Vangel, M., Banks, D. L., Heckert, N. A., Dray, J. F., & Vo, S.
- Affiliations
- National Institute of Standards and Technology (US)
- Venue
- NIST Special Publication 800-22 Revision 1a
- Published
- 2010
- OpenAlex
- W1517403092
- Semantic Scholar
- 961fc0f6c68303cde94a28fdb24e6eaf3ef040b4
- Cited by
- 3,617
- References
- 34
- Citations by year
- 2026: 128 · 2025: 205 · 2024: 192 · 2023: 208 · 2022: 166 · 2021: 291 · 2020: 290 · 2019: 250 · 2018: 254 · 2017: 207 · 2016: 183 · 2015: 184 · 2014: 201 · 2013: 155 · 2012: 139
- Type
- report
- Language
- en
- Open access
- gold
- doi.org
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Cryptographic Implementations and Security
- Physical Sciences › Mathematics › Discrete Mathematics and Combinatorics › Coding theory and cryptography
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Chaos-based Image/Signal Encryption
- Keywords
- random number generators, pseudorandom number generators, cryptographic randomness, NIST SP 800-22, randomness testing
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-09-27 (prompt extract-v2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: A statistical test suite for random and pseudorandom number generators for cryptographic applications. https://app.cochat.ai/papers/10.6028/nist.sp.800-22r1a
- The paper
- Bassham, L. E., Rukhin, A. L., Soto, J., Nechvatal, J. R., Smid, M. E., Barker, E. B., Leigh, S. D., Levenson, M., Vangel, M., Banks, D. L., Heckert, N. A., Dray, J. F., & Vo, S. (2010). A statistical test suite for random and pseudorandom number generators for cryptographic applications. NIST Special Publication 800-22 Revision 1a. https://doi.org/10.6028/nist.sp.800-22r1a
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.






