The Most Referenced Verses of the Bible

Search for “most popular Bible verses” and you will find hundreds of lists. Almost all of them are one writer’s opinion, or one website’s traffic, presented without a method. Many quietly copy each other. We wanted something we could defend: a ranking where every number traces to a published, citable source, and where no single list - however large - can dominate the outcome.

That is why this report measures references rather than popularity. Popularity is a feeling; references can be counted. A verse ranks high here only if it is referenced widely (by many sources), prominently (near the top of ranked lists), and across different kinds of evidence (read and preached and memorized, not merely anthologized). Fifteen blogs repeating each other is one piece of evidence fifteen times; our method is built so it counts that way.

The project began as a simple cross-reference of fourteen published lists and grew, over several research passes, into a 52-source study. Each pass widened the evidence: we captured the major Christian sites’ curated lists, then the behavioral data (year-end reports from Bible Gateway and YouVersion going back more than a decade, the complete 1,000-verse TopVerses search ranking, Kindle highlighting data, tweet-volume studies), then the institutional memorization canons, and finally the lone published sermon-frequency corpus.

Every citation was normalized to the individual-verse level. Published lists are messy: one cites “Psalm 23,” another “Psalm 23:4,” a third “Psalm 23:1–6.” We expanded every range and every whole-chapter citation so that all three count toward the same verses, using a standard Bible versification library - 4,092citations parsed with zero errors. Where one outlet republished another’s data (a news story covering a Bible Gateway year-end report, for instance), we kept only the original, so the same measurement is never counted twice.

The process caught real errors in the raw material. The 1,000-verse search ranking had a twenty-place numbering shift partway through its list, which we detected by re-scraping overlapping pages and realigning every rank below the gap. One AI model’s “non-repetitive” 1,000-verse list repeated a verse across its two tiers, which deduplication caught. And where a source could not be captured cleanly - a handful of magazine sites block all automated readers, and forum data was out of reach for this phase - we left it out rather than estimate. No number in this report is imputed or guessed.

Each verse is scored on three axes, blended 60/20/20:

John 3:16 is in a class of its own. Its score (66.8) clears second place by more than fifteen points - the largest gap anywhere in the top 100. It is the most-preached verse in the 91,000-sermon corpus, the most-searched verse on Google by a factor of more than twenty, and the only top-ten verse cited by all six categories of evidence at once.

Ten verses are referenced by every kind of evidence we measured: John 3:16, Matthew 28:19, Galatians 5:22, Ephesians 2:8, John 14:6, Galatians 5:23, Romans 5:8, Romans 12:1, Acts 1:8, John 1:1. These are the verses with the broadest possible footprint - read, searched, preached, cross-referenced, anthologized, and memorized. It is a striking list: alongside the expected John 3:16 sit the Great Commission, the fruit of the Spirit, and the grace-through-faith summary of Ephesians 2:8.

Breadth and depth tell different stories. Philippians 4:13 is cited by more sources than any other verse (39 of 52), with Jeremiah 29:11 and Proverbs 3:5 each at 38 - all three ahead of John 3:16’s 36. Yet all three trail it in the final ranking, because John 3:16 sits higher on the lists that carry the most weight. Being cited everywhere is not the same as being cited first.

What people lean on privately is not what editors showcase. Isaiah 41:10 (“Fear thou not; for I am with thee”) was YouVersion’s global Verse of the Year five times between 2018 and 2025 - no other verse comes close - and ranks second in app engagement lists year after year. But it rarely cracks the top ten of editorial round-ups. The gap between the usage data and the editorial lists is one of the most interesting patterns in the dataset, and verses of comfort in difficulty account for most of it.

Whole chapters behave like single verses.Psalm 23 and Psalm 91 dominate Bible Gateway’s recent year-end data verse by verse - in 2025, the top six most-read verses were the six verses of Psalm 23, in order. People do not read these psalms a verse at a time; they return to them whole. Verse-level rankings necessarily split that devotion across many entries, which is why chapter-level analysis appears alongside the verse ranking in our dataset.

The long tail is real, and we kept it visible. 190 verses are cited by five or more sources; 78 by ten or more; and 1,315 of 2,085 appear in only one source, almost all from the deep tails of the two 1,000-verse lists. Rather than trimming the tail to make the list look tidier, we flag single-source verses in the table below - the gradient from consensus to obscurity is itself part of the finding.

A ranking is only as good as its caveats. The sermon corpus covers a single year (2021), and it is the only verse-level sermon dataset ever published - so it carries a deliberately small weight, as does the single scholarly cross-reference source. Several behavioral sources share platforms across years (Bible Gateway and YouVersion each appear multiple times); the category-averaging design prevents any one platform from dominating, but those observations are not fully independent. The AI baselines cannot cite their own derivation, which is why they carry the smallest weight of all. And the underlying data skews toward English-language, online Bible engagement - the verses most referenced in Amharic preaching or Mandarin Bible study may rank differently.

The planned second phase adds the signal this dataset most wants: organic citation counts - how often each verse is actually quoted in public forums, social conversation, and sermon transcripts in the wild, counted directly rather than taken from published lists. The scoring model already has a place reserved for it as a seventh evidence category, and because every step of this study is scripted, the entire ranking can be recomputed the day that data lands.