What is the definitive Cyrillic transliteration table for converting Russian to English?

Converting Russian Cyrillic text into the Latin alphabet is a deceptively simple task that hides a surprising amount of complexity. There is no single universal table; instead, there are several competing standards, each designed for a different purpose. This guide lays out the main systems, gives you the full letter-by-letter tables, explains where they disagree, and helps you pick the right one for your project — whether you are transliterating a passport, a bibliographic citation, a genealogical record, or a large corpus of documents.

The Direct Answer: Which Table Should You Use?

Also worth reading: What are the Russian passport name transliteration rules, and how do they affect translated documents? · How accurate is AI Russian to English translation in 2026? · How do I translate English to Russian with AI in 2026?

For most modern purposes, the answer is one of three systems. If you need an official, government-recognized standard, use GOST R 52535.1-2006 (the current Russian state standard, aligned with ISO) or ICAO Doc 9303 for machine-readable travel documents. If you are working in academic publishing or library science, use the ALA-LC romanization table (American Library Association / Library of Congress), which is the dominant standard in English-language catalogs. If you want maximum linguistic precision and reversibility, use the scientific transliteration system (also called the scholarly or academic system), which is based on ISO 9 and maps every Cyrillic letter to exactly one Latin character.

The practical reality is that everyday usage on the internet follows none of these strictly. Most Russians typing in Latin script use a loose phonetic convention called translit (транслит), which produces spellings like "privet" instead of "privyet" and "zhizn" without any diacritics. So when someone asks for "the" Cyrillic transliteration table, the honest answer is: it depends on whether you need officialdom, scholarship, or casual readability. Below we give all of them.

The Core Russian Alphabet and Its Latin Equivalents

The modern Russian alphabet contains 33 letters. Twenty of them map cleanly to single Latin letters across nearly every standard, which is why the core of any transliteration table looks similar no matter which system you choose. The disagreements concentrate in about ten letters, mostly those representing sounds English lacks as single letters: the palatalized consonants, the affricates, and the vowels я, ё, ю.

Here is the consensus core, valid under ALA-LC, GOST, ISO 9, BGN/PCGN, and scientific transliteration alike:

That covers twenty letters. The remaining thirteen — Е, Ё, Ж, Й, Х, Ц, Ч, Ш, Щ, Ъ, Ь, Ю, Я — are where the standards diverge, sometimes dramatically.

Where the Standards Disagree: Full Comparison Table

The following table compares the five major systems on the contested letters. Note how ISO 9 uses diacritics to guarantee one-to-one reversibility, while BGN/PCGN and GOST favor digraphs that read more naturally to English speakers.

A worked example makes the differences concrete. The surname Щербаков renders as Ščerbakov in scientific transliteration, Shcherbakov in ALA-LC and BGN/PCGN, Shcherbakov under GOST, and anything from Scherbakov to Stcherbakoff in older or casual sources. The name Юрий becomes Ûrij, Iurii, Yuriy, Yuriy, or simply Yuri depending on the system. None is wrong; each serves its audience differently.

The Major Systems Explained One by One

Scientific transliteration (ISO 9). This system assigns one Latin character to each Cyrillic character, using diacritics (ž, č, š, ŝ, â, û, ë) to cover the surplus. Its great virtue is perfect reversibility: you can convert Latin back to Cyrillic with zero ambiguity, because no two Cyrillic letters ever share a Latin rendering. Its weakness is readability for non-specialists — few English readers know that "ŝ" is pronounced "shch." It dominates Slavic linguistics journals and encyclopedic work in continental Europe.

ALA-LC. Used by virtually every major research library in North America, this system favors digraphs (sh, ch, zh, shch, iu, ia) over diacritics, with only ĭ for short i and apostrophes for the hard and soft signs. It is not fully reversible — "ia" could come from either я or the letter sequence и-а — but catalogers accept this trade-off for readability. If you cite Russian books in an English-language bibliography, this is almost certainly the standard your publisher expects.

BGN/PCGN. Developed jointly by the US Board on Geographic Names and the Permanent Committee on Geographical Names for British Official Use, this system was built for maps and place names. It reads smoothly for English speakers (ya, yu, yo, kh, ts) and remains the basis for many geographic databases. It allows some context-dependent rules, such as "ye" at word beginnings versus "e" elsewhere, which slightly complicates automated processing.

GOST R 52535.1-2006. The current Russian federal standard, adopted in 2006 and harmonized with international practice for documents like passports. An earlier standard, GOST 16876-71, used different renderings (j for й, c for ц) and still appears in older Soviet-era materials. GOST's handling of the soft sign is unusual: it transliterates ь as "ie" after consonants at stem endings, producing forms like Tatiana from Татьяна rather than Tat'iana.

ICAO Doc 9303. For machine-readable passports worldwide, ICAO mandates a simplified scheme: е → e, ё → e, й → i, х → kh, ц → ts, щ → shch, ъ and ь omitted, ю → u, я → a. This is why a Russian passport shows Yuriy as "Yuri" and Pyotr as "Petr." The scheme sacrifices phonetic detail for consistency across borders.

How to Transliterate Text Step by Step

For a manual conversion of a short text, follow this sequence. First, identify the source standard if one exists — a passport, a published book, or an official form may already dictate the system, and overriding it creates mismatches with other records. Second, choose your target standard deliberately and note it in your output; mixing systems within one document is the most common quality failure. Third, process the text letter by letter using your chosen table, watching for the context-dependent rules: BGN/PCGN's ye/e split, GOST's soft-sign handling, and the treatment of pre-reform orthography if you are working with texts printed before the 1918 spelling reform (which contained the letters і, ѣ, ѳ, and ѵ).

Fourth, handle proper nouns with care. Established English exonyms override mechanical transliteration: Москва is "Moscow," not "Moskva," and Санкт-Петербург is "Saint Petersburg" even though BGN/PCGN would yield "Sankt-Peterburg." Fifth, verify reversibility requirements before you start. If the Latin text must be converted back to Cyrillic later — for example, in a database migration — only ISO 9 guarantees lossless round-tripping, and digraph-based systems will require disambiguation logic.

Common Mistakes and How to Avoid Them

The most frequent error is treating transliteration as translation. Transliteration converts script; translation converts meaning. Confusing them leads people to ask for the "English version" of a name when what they actually need varies by context: a visa application needs ICAO-style transliteration matching the passport, while a marketing document might legitimately translate a company name.

Second, people ignore the letter ё. Although officially part of the alphabet since 1942-era standardization debates and always present in learner materials, ё is routinely printed as е in Russian publications, so the same name can appear as both Semyon and Semen, Pyotr and Petr. Decide on a policy: either preserve ё when present in the source, or normalize everything to е and accept the ambiguity.

Third, apostrophes and quotation marks get lost in data pipelines. ALA-LC uses straight apostrophes for ь and double quotes for ъ; HTML encoding, CSV exports, and database collations frequently strip or mangle these characters. Fourth, older anglicizations persist misleadingly: Tchaikovsky (from German-influenced Tschaikowsky), Tolstoy (versus Tolstoi), Dostoevsky (versus Dostoevskii). These are conventional English spellings, not errors, but they should never be mixed with systematic transliterations in the same document. Fifth, Ukrainian, Belarusian, Bulgarian, and Serbian Cyrillic have their own distinct transliteration standards — applying the Russian table to Ukrainian names, for instance, produces incorrect results for letters like і, ї, and ґ.

Manual vs. Automated Transliteration

For a handful of names, manual conversion with a printed table takes minutes and gives you full control over edge cases. For bulk work — thousands of records, scanned archives, ongoing feeds — automation is necessary, and here the choice of standard matters enormously. Context-independent systems like ISO 9 can be implemented as a simple character mapping table and run at millions of characters per second with zero ambiguity. Digraph systems require lookahead logic (is "e" word-initial? does "ie" represent ь or a genuine vowel pair?), which raises error rates unless carefully engineered.

Modern AI-based approaches add another layer: rather than mapping characters mechanically, neural models can learn from context, resolving ambiguities like ё/е normalization, recognizing established English spellings of famous names, and handling mixed-script input. Services such as AI Translations apply this kind of contextual processing when converting between scripts, which matters most for messy real-world data — OCR output from old documents, informal translit typed by users, or multilingual corpora where Russian appears alongside Ukrainian and Belarusian. The trade-off is transparency: a deterministic table always behaves predictably, while a statistical model occasionally surprises you, so production systems should log their conversions for auditability.

Cost-wise, manual transliteration is free but labor-intensive; rule-based software ranges from free open-source libraries to commercial tools priced per volume; AI-powered services typically charge per character or via monthly plans, with rates varying widely by provider and volume. For a one-off project of under a few hundred names, manual work with a reliable table is usually the sensible choice regardless of budget.

When Each Standard Applies: Practical Decision Guide

Use ICAO Doc 9303 whenever a document must match a passport or official ID — visas, banking KYC checks, airline bookings. Airlines reject tickets whose passenger names do not match the machine-readable zone, and the mismatch rate between casual transliteration and passport spelling is high enough to cause real problems at check-in.

Use ALA-LC for bibliographies, library cataloging, academic citations, and archival finding aids in North America. Use ISO 9 / scientific transliteration for linguistics papers, etymological discussion, and any dataset that must be reversible. Use BGN/PCGN for geographic features, maps, and gazetteer entries. Use GOST R 52535.1-2006 for documents governed by Russian regulation. And use plain casual translit only for informal communication, never for records.

Timing matters too: if you are building a system that will store names long-term, decide the standard now and enforce it from day one. Retrofitting transliteration standards across an existing database is far costlier than choosing correctly upfront, because deduplication fails silently — "Tatyana Ivanova" and "Tat'iana Ivanova" look like different people to a naive string match, and merging those records later requires fuzzy-matching infrastructure that a consistent initial policy would have made unnecessary.

Historical Notes Worth Knowing

Transliteration standards have shifted repeatedly, and awareness of this prevents confusion with older sources. Before roughly the 1990s, French and German conventions heavily influenced English renderings of Russian: "ou" for у (as in "Kerensky" era spellings like "Tolstoï"), "kh" competing with German "ch," and feminine surname endings appearing as "-off" or "-eff" instead of "-ov/-ev." The Soviet Union standardized internal romanization through successive GOST editions (1938, 1947, 1956, 1968, 1971, 1983, 1986), each with slightly different tables, so a 1960s atlas and a 1980s encyclopedia may romanize the same place name differently.

The Cyrillic alphabet itself descends from the ninth-century Glagolitic and Cyrillic scripts created for Old Church Slavonic, traditionally credited to the disciples of Saints Cyril and Methodius. The 1918 Bolshevik spelling reform dropped four letters (і, ѣ, ѳ, ѵ) and eliminated the silent hard sign at word ends, which is why pre-revolutionary texts end words with ъ and contain spellings like съездъ. Anyone transliterating historical documents must first determine which orthographic era the source belongs to before applying any modern table.