Six in ten OFAC SDN entries were added after 2021, and nearly half have no alias

OFAC SDN list statistics: how many names are on the list?

The figure you get depends on what you count, which is why pages that rank for this question quote anything from about 8,000 to more than 20,000, usually without a date or a unit [7][8][9][11]. We count entries: one record per sanctioned person, company, vessel or aircraft, each with its own OFAC unique ID. On 5 October 2026 the SDN list had 19,363 entries, the list our OFAC screening API screens against. OFAC's legacy SDN.XML file published on the same day states the same record count, and contains the same number of records, so the two official formats agree.

Entities are the largest group at 9,990 entries (51.6%), followed by 7,492 individuals (38.7%), 1,539 vessels (7.9%) and 342 aircraft (1.8%). Another 481 records sit only on OFAC's non-SDN lists, such as the Sectoral Sanctions Identifications list (that file's data date is 14 September 2026), and 93 SDN entries also appear on one of those lists. Those non-SDN records are outside this study; see how the main sanctions lists differ.

Counting names rather than entries gives a different answer. The same 19,363 entries carry 24,596 aliases, so a screening system that indexes every published name works with 43,959 Latin-script names before it adds any spelling variants of its own.

How old are the entries on the SDN list?

Each entry in the enhanced XML carries the date it was published to the SDN list. We checked that field against OFAC's own change files [5]. We hold 170 of the 520 delta files OFAC has published since September 2022, covering 164 publication dates. For all 4,002 additions in those files that are still listed, the date in the list file matches the date of the delta file exactly. So the field records when an entry joined the list, not when it was last edited.

Most of the list is recent. 11,815 entries (61.0%) were first published on or after 1 January 2022, and 11,710 (60.5%) on or after 24 February 2022. The median entry was published on 8 December 2022. Only 1,390 entries (7.2%) predate 2010, and the oldest still listed goes back to January 1981.

The share is higher for some types: 74.8% of vessels and 68.0% of entities were added since 2022, against 50.3% of individuals. This is a survivor count. Entries that were added and later removed do not appear, so the chart shows the age of today's list, not the history of designations. That history is the subject of a separate study, how often the OFAC SDN list changes.

Which sanctions programmes make up the SDN list?

OFAC tags each entry with one or more programme codes, the legal authority it was designated under. The list uses 72 distinct tags. One of them dominates: RUSSIA-EO14024, the Russia programme under Executive Order 14024, covers 6,346 entries (32.8%). Counter-terrorism (SDGT) follows with 3,347 (17.3%), then the Iranian Financial Sanctions Regulations (IFSR) with 1,620 (8.4%) and the narcotics Kingpin Act (SDNTK) with 1,363 (7.0%).

Entries are not neatly filed under one programme. 3,016 entries (15.6%) carry two or more tags, and one carries eight. Reports that sum entries by programme therefore overcount the list, and the shares in Figure 3 add up to more than 100%.

The programme mix matters for matching because programmes differ in how names are published. Entries under SDGT average 2.42 aliases each, while entries under IRAN-EO13902 average 0.41 and 73.5% of them have none.

How many aliases does an SDN entry have?

An alias (OFAC writes a.k.a., f.k.a. or n.k.a.) is any other name OFAC publishes for the same target. Aliases are where a lot of screening effort goes, so it is worth knowing how unevenly they are spread. The mean is 1.27 aliases per entry, but the median is 1 and the most common value is zero. 9,399 entries (48.5%) have no alias at all.

For people and companies the pattern is close: 43.0% of individuals and 44.8% of entities have no alias, about a quarter have exactly one, and only 5.5% of individuals and 3.8% of entities have six or more. Vessels and aircraft are almost alias-free (88.6% and 99.4% have none) and carry registration identifiers instead. The long tail is real but small: the largest alias set is 95 aliases on one counter-terrorism (SDGT) entity.

Nearly half of the list (48.5%) gives a matching engine only the primary name, in Latin script and sometimes in one other script. Whether an engine catches a customer whose name is a different transliteration of that name is not something the list will help with.

How many SDN aliases are weak?

Since 2011 OFAC has flagged some aliases as weak [10]. In its words, a weak AKA is “a relatively broad or generic alias that may generate a large volume of false hits” [2], one source of false positives in sanctions screening. OFAC also says it “does not expect that persons will screen for weak AKAs”, but that they may be used to confirm a hit raised by other information [3]. The enhanced XML marks these with a low-quality flag on each alias.

4,417 of the 24,596 aliases (18.0%) carry that flag. The share is higher for individuals (2,266 of 10,891 aliases, 20.8%) than for entities (2,080 of 13,420, 15.5%), and highest for vessels (71 of 283, 25.1%).

Put the two facts together and the list looks thinner still. 1,130 entries (5.8%) have aliases, but only weak ones. If a firm follows OFAC's general expectation and does not screen weak aliases, those entries are matched on the primary name alone, exactly like the 9,399 entries with no alias. That is 10,529 entries (54.4%) with a single name to screen. Counting the primary name and strong aliases only, the median entry has one screenable name and the mean is 2.04.

Which identifiers does OFAC publish for listed people?

Identifiers are what turn a name hit into a decision. For the 7,492 individuals on the list, the picture is good on birth dates and weaker on documents. 7,421 (99.1%) have some birth date, and 6,527 (87.1%) have at least one exact date with day, month and year. 532 (7.1%) have two or more birth dates, which a matching rule that expects one date will mishandle. 6,073 (81.1%) have a nationality or citizenship, 4,795 (64.0%) a place of birth and 5,540 (73.9%) a gender.

Documents are the gap. 2,085 individuals (27.8%) have a passport number and 3,843 (51.3%) have a passport or a national ID-type number. 875 individuals (11.7%) have neither an ID number nor a nationality, which leaves the name and any birth date or address OFAC publishes.

The age of an entry predicts its quality. OFAC said in July 2026 that older entries “sometimes lack identifying information that is now routine for new sanctions actions” [4], and the file bears this out. Individuals added since 2022 almost always carry an exact birth date (94.3%), a nationality or citizenship (99.3%) and a gender (99.9%). Individuals added before 2010 carry them 76.0%, 51.7% and 8.6% of the time. One thing has not improved since 2010: the share with a passport or national ID is 52.8% for both 2010 to 2021 and 2022 onward.

Entities and vessels fare better on registration data: 7,981 of 9,990 entities (79.9%) have a registration or tax number, and 1,529 of 1,539 vessels (99.4%) have a vessel registration identification number.

How often do SDN entries share a name?

Two different SDN entries with the same name are a problem for any process that stops at the first match. Exact duplicates of primary names are rare. After normalising case, accents and punctuation (and ignoring word order for people), 54 entries in 26 groups share a primary name with another entry of the same type: 24 entities, 16 individuals and 14 vessels. For 7 of the 8 groups of same-named individuals, the published birth years differ, so the birth date is what separates them.

Counting aliases as well, 338 entries (1.7%) share at least one identical name with another entry of the same type, 189 of them individuals.

Near-identical names are a different story for vessels and aircraft. 417 of 1,539 vessels (27.1%) and 303 of 342 aircraft (88.6%) have a primary name one character away from another entry's. In these two groups many names differ from another listed name by a single digit or letter, so a fuzzy name match on a vessel or aircraft says little on its own. For people (60 of 7,492, 0.8%) and companies (115 of 9,990, 1.2%) the rate is low.

4,840 entries (25.0%) include at least one name in a non-Latin script published by OFAC: Cyrillic for 2,928 entries, Arabic for 1,334 and Chinese (simplified or traditional characters) for 517. Whether a name is of Arabic, Persian or Russian origin cannot be read from the file reliably, so we report published scripts, not origins.

What this means for compliance teams

1. Check how your matching handles entries with one name. Half of the list (48.5%) has no alias, and 54.4% has no strong alias. Test whether your system still finds those entries when the customer name is a different transliteration, word order or spelling of the single published name. Our alias screening benchmark shows one way to measure this on real OFAC names, and the vendor test kit lists the cases to try.
2. Decide, in writing, what you do with weak aliases. 18.0% of aliases are flagged weak. OFAC's FAQ says it does not generally expect firms to screen them, so your policy should state whether you screen them, suppress them, or use them only to confirm a hit, and your test set should prove the system does what the policy says. Our matching methodology shows how Verifex weighs aliases with other identity evidence.
3. Use birth dates, and handle more than one. 87.1% of listed individuals have an exact birth date, and 7.1% have several. Check that your rules compare every published date, not only the first.
4. Do not expect documents to save you. Only 51.3% of listed individuals have a passport or national ID number. A process that relies on document numbers to clear hits will work for about half of individual entries.
5. Treat vessels and aircraft as identifier problems, not name problems. 27.1% of vessel names are one character from another listed vessel. Screen vessel registration identifiers, which 99.4% of listed vessels carry.

Methods

Data. OFAC Sanctions List Service files for the SDN list, downloaded on 6 October 2026 between 21:36 and 21:37 UTC [1]. The file header gives the data date as 5 October 2026. Our data sources page lists the lists Verifex loads.

Processing. A Python script reads the enhanced XML and writes one record per entry with its type, list memberships and SDN publication date, programme tags, every name (primary or alias, alias type, OFAC's low-quality flag, script and full formatted name), features (birth dates with exact, approximate or range flags, nationality, citizenship, place of birth, gender) and identity documents. A second script computes every figure in this study and prints it.

Definitions. An entry is one record with its own OFAC unique ID. An alias is any non-primary name on the entry (a.k.a., f.k.a. or n.k.a.), counted once whatever number of script translations it has. A weak alias is an alias with isLowQuality set to true in the enhanced XML [6], which is how OFAC marks weak AKAs [10]. An exact birth date has day, month and year and is neither approximate nor a range. Passport or national ID means a document of type Passport, Diplomatic Passport, or a national identity number type (for example National ID No., Personal ID Card, Cedula No., D.N.I., C.U.R.P., Tazkira). An identical name is equal after upper-casing, removing accents, apostrophes and other punctuation, and, for individuals, sorting the words. A one character edit is a Levenshtein distance of one between normalised primary names of the same entry type.

Checks. The entry count was confirmed against the Record_Count field and the number of records in SDN.XML (19,363 in all three). The largest programme count was recomputed by direct filtering, and the alias partition (none, weak only, at least one strong) sums to the total. The meaning of the publication date was validated against the 170 OFAC delta files we hold: 4,002 of 4,002 additions in them that are still listed match the date in the list file.

Statistics. Every number is a full count of the list at the snapshot, not a sample, so we report no confidence intervals.

This study describes public data and general regulatory context. It is not legal advice; confirm your obligations with a qualified adviser.

Limitations

• The study describes one snapshot (data as of 5 October 2026). OFAC changes the list most weeks, so counts drift.
• Publication dates describe today's entries only. Entries removed before the snapshot are not counted, so Figure 2 understates designations in earlier years.
• Identifier completeness counts what OFAC publishes, not whether it is correct. A wrong birth date counts as present.
• The national ID category relies on OFAC's document type labels. Some document types are ambiguous (for example “Identification Number”), so the national ID share could move by a few points under a different rule.
• Name origin (Arabic, Persian, Russian and so on) cannot be inferred from the file without guessing, so we report only the scripts OFAC publishes. The share of names of non-Latin origin is not available from public data; answering it would need a labelled name-origin dataset.
• Name collision counts depend on the normalisation used. Phonetic or transliteration-aware comparison would find more near matches.
• The study does not measure how any screening system performs on these entries.

Questions and answers

Data and code

Every value plotted in Figures 1 to 7 is in the figures file below, and the headline counts are in the summary file. Both are aggregates computed from OFAC's public SDN files; they contain no personal data beyond what OFAC publishes, and no customer data was used. The source files and their SHA-256 hashes are listed under Methods, so anyone can repeat the count.

• jsonFigures 1 to 7: every plotted value
• jsonHeadline counts