Resources & Analytics

XRPL TPS, Measured: What 13 Years of Ledger Data Actually Show

How fast does the XRP Ledger really run? We analyzed every ledger from 2013 to September 2025 to measure XRPL throughput honestly — and what we found differs from the headline numbers.

XRPLWin Team
· 18 min read
XRPL TPS, Measured: What 13 Years of Ledger Data Actually Show

How fast does the XRP Ledger actually run? It's a question that has been answered loosely for over a decade — by marketing materials, comparison tables, comment threads, and eyeballed estimates. We wanted to answer it precisely, using the only source that can't be argued with: the network's own validated ledger record from launch in 2012 through September 2025.

The answer is more interesting than any single number. Routine sustained throughput on the XRPL has climbed from essentially nothing in 2013 to a typical daily peak of 123.5 TPS in 2025 — a real, gradual transformation. The all-time peak sustained throughput is 228 TPS over a 30-second window, recorded on July 21, 2025. The network has never sustained 1,000 TPS over any meaningful wall-clock window in its entire history.

That last fact runs counter to many published claims about XRPL. So the rest of this article walks through how we measured, what we found, and where the confusion comes from.

A note on what "TPS" means

This matters before any number makes sense.

The XRP Ledger records ledger close times at 10-second resolution (per the protocol documentation) — close times round to values ending in :00, :01, :02, :10, :11, :20, and so on, with adjustments to keep them strictly increasing. The actual wall-clock interval between consecutive ledgers averages about 3.9 seconds. This is a deliberate design choice: 10-second rounding makes consensus easier to reach across geographically distributed validators.

A consequence: when one ledger has a recorded close time of 12:00:01 and the next has 12:00:02, the recorded delta is 1 second — but the real wall-clock interval might have been 4 seconds. Two ledgers that genuinely closed 4 seconds apart may show a 1-second recorded delta, or a 10-second one, depending on how rounding fell.

This affects how TPS is calculated. There are two common conventions:

Per-ledger TPS — the formula used by major blockchain explorers (Etherscan, Solscan, others): transactions_in_ledger ÷ recorded_close_time_delta. Easy to compute, useful for spotting transaction-density peaks, but inflated when consecutive ledgers happen to share short recorded deltas.

Wall-clock sustained TPS — total transactions across a fixed time window (e.g., 30 or 60 seconds), divided by the window length. Smooths over close-time rounding artifacts. This is what most people actually mean when they ask "how fast is the network."

Both measurements are real. Both have legitimate uses. But they give very different numbers, and conflating them has produced a decade of confused throughput discussions about the XRPL. Throughout this article, when we say "TPS" without qualification, we mean wall-clock sustained TPS over a 30-second window — the more honest of the two.

How we measured

Our indexer continuously ingests validated ledgers and transactions directly from the XRPL network and stores them in Indexed Database, optimized for analytical queries across the full ledger history. Coverage runs from December 31, 2012 through September 4, 2025.

For wall-clock TPS, we group ledgers into 30-second buckets, sum the transactions in each bucket, and divide by 30 to get a real per-second rate. The maximum value across a day becomes that day's peak. This neutralizes the close-time rounding issue entirely — across 30 seconds of wall-clock time, individual ledger rounding artifacts average out.

For per-ledger TPS (used in some comparison points below), we use the standard explorer formula: transactions in a ledger divided by the recorded close-time delta to the previous validated ledger.

All figures are derived from validated mainnet ledgers; no testnet data is included.

13-year throughput summary

The table below shows what the network actually did, year by year. Daily peak TPS is the highest 30-second sustained rate observed each day, then aggregated across the year. Daily average TPS is total daily transactions ÷ 86,400 seconds.

  • 2013 — median daily peak 1.4, yearly max 60.5, daily avg 0.34, days above 50 TPS: 2
  • 2014 — median 9.8, yearly max 95.0, daily avg 2.25, days above 50 TPS: 6
  • 2015 — median 20.9, yearly max 66.7, daily avg 3.24, days above 50 TPS: 7
  • 2016 — median 19.6, yearly max 134.5, daily avg 7.39, days above 50 TPS: 2, days above 100: 1
  • 2017 — median 20.9, yearly max 57.1, daily avg 8.97, days above 50 TPS: 2
  • 2018 — median 18.2, yearly max 47.5, daily avg 7.83, days above 50 TPS: 0
  • 2019 — median 21.7, yearly max 148.1, daily avg 11.16, days above 50 TPS: 34, days above 100: 10
  • 2020 — median 26.1, yearly max 105.8, daily avg 11.84, days above 50 TPS: 26, days above 100: 1
  • 2021 — median 25.9, yearly max 84.9, daily avg 14.01, days above 50 TPS: 24
  • 2022 — median 28.4, yearly max 195.2, daily avg 15.21, days above 50 TPS: 28, days above 100: 7, days above 150: 3
  • 2023 — median 30.5, yearly max 125.8, daily avg 14.00, days above 50 TPS: 20, days above 100: 4
  • 2024 — median 39.1, yearly max 181.1, daily avg 22.21, days above 50 TPS: 136, days above 100: 29, days above 150: 5
  • 2025 (through Sep 4) — median 123.5, yearly max 228.3, daily avg 22.91, days above 50 TPS: 246, days above 100: 177, days above 150: 69

Three facts jump out.

First, the network's all-time peak sustained throughput is 228 TPS over a 30-second window, recorded on July 21, 2025. Not 500. Not 1,500. Two hundred twenty-eight.

Second, the 2022 peak (195 TPS sustained) is only 33 TPS below the all-time record. The single highest-TPS day in XRPL history before 2025 was June 23, 2022 — a fact rarely cited in conversations about XRPL performance.

Third, 2025 represents a genuine structural shift in routine throughput, not a peak record. The median daily peak in 2025 is 123.5 TPS — over 3× the 2024 figure (39.1) and over 4× any prior year. 246 of 247 days in 2025 saw peaks above 50 TPS, against 136 days in all of 2024 and zero days in 2018. The peak number didn't change much. The floor moved.

Three eras visible in the data

The 13-year history breaks down into three distinct phases.

Bootstrap (2013–2015): Median daily peak under 25 TPS. The infrastructure was operational; the demand was not yet there. Yearly maximum touched 95 TPS once in 2014 and 67 TPS in 2015, but routine activity stayed below 5 TPS in average terms. The network was waiting for users.

The middle era (2016–2023): Daily peaks settled into a 18–30 TPS routine pattern. Yearly maxima fluctuated — 134 TPS in 2016, 195 TPS in 2022, 126 TPS in 2023 — depending on whether any individual day produced a notable burst. The 2017–2018 crypto bull market is conspicuously absent from this view: 2018 saw zero days above 50 TPS, even as XRP hit its first all-time high. Volume was growing in price terms; transaction throughput was not. Within this era, two periods stand out:

  • 2019 — Despite crypto markets in deep bear territory, 2019 saw 34 days above 50 TPS and 10 days above 100 TPS, both more than any prior year. No major announcement accompanied this; it shows up only in the indexed data.
  • June 2022 — A multi-day cluster of high-throughput activity peaked on June 23 at 195 TPS sustained. This was the highest-throughput day in XRPL history until 2025, but it was nearly invisible publicly because the broader crypto market was reeling from the LUNA collapse.

The modern era (2024–2025): Median daily peak rose to 39 TPS in 2024 and to 123 TPS in 2025. Days above 100 TPS climbed from 29 to 177 in eight months. This shift is real and measurable, and it coincides with structural changes the network deserves credit for: the AMM amendment, the substantial user-base expansion of late 2024, the RLUSD launch, and the operational maturity of the surrounding ecosystem.

What 1,500 TPS actually was

The number originated in a September 2017 Ripple blog post titled "The Most (Demonstrably) Scalable Blockchain", which stated that the XRP Ledger could sustain 1,500 transactions per second on commodity hardware. This followed an earlier March 2017 announcement of internal benchmarks across 16 geographically distributed validators reaching nearly 1,000 TPS. Both were testbed results — not measurements of the live network.

In April 2023, Ripple's CTO Emeritus David Schwartz responded to a community member quoting the 1,500 figure: the team had never seen 1,500 TPS on the live XRPL, and the network as configured "probably sustains 300 to 500." The original phrasing on the website, he said, was poorly chosen — it was meant to communicate what the technology could handle, not what the live network was doing. In May 2026, Schwartz weighed in again, saying the technology could ramp to 1,500 TPS under realistic conditions but the live deployed network could not.

Looking at our data, even Schwartz's "300 to 500 TPS" estimate appears generous. Under wall-clock methodology, the live XRPL has never sustained more than ~230 TPS over any 30-second window in its entire history. The architecture may be capable of higher; the production network has not demonstrated it.

A different number worth knowing: peak per-ledger transaction density

On February 21, 2025, ledger 94303842 closed on the XRP Ledger carrying 1,945 transactions. Its recorded close time was exactly 1 second after the previous ledger's. Under the per-ledger TPS formula (transactions ÷ recorded close-time delta), this works out to "1,945 TPS."

That figure is real, in a specific sense: 1,945 transactions did go into a single ledger that recorded a 1-second delta to its parent. But the actual wall-clock interval between those two ledgers was almost certainly closer to 4 seconds (the network average), making the wall-clock equivalent ~486 transactions per second. And the surrounding context — over what 30 seconds the network was running — is the 192 TPS sustained figure that appears in our wall-clock peak windows for January 17, 2025.

This same measurement convention applied to XRPL's earlier history shows ledgers in 2022 and 2024 hitting 1,000+ TPS by the per-ledger formula. It's a real metric of transaction density per consensus round. It is not a metric of wall-clock throughput.

What February 2025 actually was

Even after the throughput numbers come down, February 2025 remains analytically interesting — for a different reason than headline TPS.

The all-time peak per-ledger densities cluster heavily in February 2025: every one of the top 25 highest-density ledgers in XRPL history occurred that month, clustered across a roughly three-week window. The pattern was sustained ceiling-testing of per-ledger transaction density, not record-breaking sustained throughput. The 30-second wall-clock peak of 228 TPS happened in July, not February.

So what was February 2025? Looking at the ledgers themselves:

  • The peak ledgers were 91.8% to 95.6% Payment, against a monthly baseline of 45.7%
  • 80.8% of payments in the all-time peak ledger 94303842 moved a non-XRP currency (issued tokens or IOUs)

The February peaks are concentrated bursts of payment activity from a small set of accounts. To find out what those accounts were doing, we looked at the transaction result codes.

What the result codes are saying

Every transaction on the XRP Ledger gets a result code. tesSUCCESS means the transaction did what it was meant to do. tecPATH_PARTIAL means a payment found a route but couldn't deliver the full requested amount because available liquidity along that route was exhausted. tecPATH_DRY means no viable path remained at all. Both tec* results are charged the network fee — the transaction made it into the validated ledger, the network processed it, and the sender paid. The payment just didn't deliver.

Across all of February 2025, network-wide:

  • tesSUCCESS: 93.54%
  • tecPATH_PARTIAL: 1.92%
  • tecPATH_DRY: 1.53%
  • All other failure codes combined: ~3%

In the six top peak ledgers, the picture inverts:

  • tesSUCCESS: 10.9–15.5%
  • tecPATH_PARTIAL: 65.6–77.5%
  • tecPATH_DRY: 7.6–18.0%

Looking at the top 15 most active accounts in the peak ledgers themselves: each one routed payments to exactly one unique destination, and zero of them succeeded. The success count for each top peak-ledger account is literally 0. Several show 100% tecPATH_PARTIAL (every payment finding a partial route, never enough liquidity to deliver in full); others show roughly 50/50 splits between tecPATH_PARTIAL and tecPATH_DRY (payments transitioning from partial fills to fully dry routes as liquidity drained completely).

Two activity layers

Layer A — visible service activity. Across all of February 2025, 17.9% of transactions carried identifying memos. Top tagged services include XPMarket (DeFi platform: ~530,000+ transactions across multiple memo variants), Xaman (the major XRPL wallet: 121,058 transactions across 23,751 unique senders), First Ledger (~504,000 transactions across 12,452 senders), xrp.cafe (NFT marketplace: ~70,000+ transactions), and Magnetic AMM (~50,000+ transactions). A second-tier of activity came from large airdrop-style campaigns: claim-style payment memos pointing to URLs accounted for nearly 2 million transactions across roughly 116 unique sender accounts. Layer A activity overwhelmingly succeeded —its baseline tesSUCCESS rate matches the network's monthly figure of 93.5%. This layer is what drives the 2025 sustained-throughput numbers.

Layer B — untagged peak-density bursts. The ledgers that set the per-ledger density records had a very different signature. Memo prevalence in those ledgers ran at 0.7–6.2%, against the 17.9% monthly baseline. Result codes flipped — 65–77% tecPATH_PARTIAL. Top accounts in these ledgers all routed to a single destination, with 0% success rates. This is the textbook on-chain signature of automated arbitrage activity caught in feedback loops against thin or rapidly-changing liquidity conditions.

The peak-density records on the XRP Ledger were set by automated trading systems running into exhausted liquidity routes, repeatedly submitting failing payments that the network nevertheless validated and processed. The bulk of February's broader activity — Xaman, XPMarket, First Ledger, xrp.cafe, Magnetic, the airdrop-style campaigns, RLUSD ambient flows — coexisted with this layer but was distinct from it.

The 2024 regime change

The 2024 regime change in transaction-type composition is worth knowing about because it underlies the 2025 throughput shift.

In 2022, OfferCreate accounted for 62.4% of all XRPL transactions and Payment for 17.5%. In 2023, the split was nearly identical at 62.2% / 23.2%. Then 2024 inverted: OfferCreate dropped to 33.9%, Payment climbed to 57.6%. That ratio held through 2025 at 31.3% / 57.3%.

In one year, Payment went from 23% to 58% of all transactions. Absolute volume tells the same story: 102 million Payment transactions in 2023 became 404 million in 2024 — a fourfold jump.

This shift coincided with two events. The first was the activation of the AMM amendment on March 22, 2024 — bringing XLS-30 Automated Market Maker functionality to mainnet. AMM-based trades absorbed 61% of all DEX activity within eight weeks; by October 2024, that share reached 90% and has held since. Some of the apparent shift from OfferCreate to Payment in 2024 reflects this activation: path-routed payments through AMM pools are recorded as Payment transactions, not OfferCreate.

The second event was a substantial expansion of the network's user base. Unique monthly senders quadrupled between October and December 2024, from 152,000 to 660,000. Several factors converged: the U.S. presidential election on November 5, 2024 set off a broader crypto rally, with XRP itself surging 420% in November alone from $0.50 to $2.63. By mid-January 2025, XRP had peaked at $3.19–$3.40, breaking its 2018 all-time high. And on December 17, 2024, Ripple's RLUSD stablecoin went live publicly on both the XRP Ledger and Ethereum.

The RLUSD launch is visible in the on-chain data with unusual clarity. Activity tied to the RLUSD issuer account stayed under 1,000 transactions per day during the second week of December as the launch approached. On December 16, the day before launch, RLUSD-touching transactions sat at 4,754. On December 17 they jumped to 64,489 — a 13.6× single-day increase. Unique senders went from 2,517 to 9,962 in the same 24 hours, and unique recipients from 14 to 570. The token's daily transaction count has not dropped below 70,000 in any single day since.

What the data settles

After 13 years of validated mainnet activity, several questions about XRPL throughput have empirical answers:

How fast does the live XRP Ledger actually run? The all-time peak sustained throughput over a 30-second window is 228 TPS, recorded July 21, 2025. The all-time peak over a 60-second window is ~158 TPS, recorded June 23, 2022. Routine operation in 2025 runs at roughly 21 TPS sustained on average, with daily peaks typically reaching 100–125 TPS.

Has the network ever sustained 1,500 TPS? No. The 2017 Ripple blog post that introduced the figure described a testbed benchmark, not a live-network capability. Schwartz has clarified this twice publicly. The on-chain record confirms it.

Is the 1,945 TPS figure that circulated in February 2025 real? It's real as a per-ledger transaction-density measurement (1,945 transactions in a single ledger with a 1-second recorded close-time delta), which is the convention used by major blockchain explorers. It is not a wall-clock TPS measurement, and reading it as one overstates real throughput by roughly 4×.

What was actually special about February 2025? Concentrated automated-trading bursts produced the highest per-ledger transaction densities in XRPL history, with peak ledgers running 65–77% tecPATH_PARTIAL failure rates — a clear signature of bot activity hitting exhausted liquidity routes. February 2025 set per-ledger density records, not sustained-throughput records.

Is something structurally different about 2025? Yes. The median daily peak TPS more than tripled from 2024 (39) to 2025 (124). Days above 100 TPS went from 29 in all of 2024 to 177 in the first 8 months of 2025. The network's typical operating regime has measurably shifted upward — driven by AMM activity, RLUSD, and the late-2024 user-base expansion.

The XRP Ledger is not running at 1,500 TPS. It has never run at 1,500 TPS. What it has done — quietly, continuously, and with little fanfare — is transform from an effectively-idle network in 2013 to one routinely sustaining 100+ TPS daily peaks in 2025. That's a meaningful, real story. It's just not the story that's been told.

Methodology and References

Methodology

WinDB indexed database of the full XRP Ledger. Coverage: December 31, 2012 through September 4, 2025.

Wall-clock TPS calculation: Ledgers are grouped into 30-second buckets by toStartOfInterval(Timestamp, INTERVAL 30 SECOND). Within each bucket, transactions are summed and divided by 30 to yield a per-second sustained throughput. Daily peak TPS is the maximum 30-second-bucket value within a given day.

Per-ledger TPS calculation (referenced for comparison): For each closed ledger, transactions are divided by the recorded close-time delta to the previous validated ledger using lagInFrame(Timestamp) OVER (ORDER BY Timestamp). This is the convention used by major blockchain explorers (Etherscan, Solscan, others). Note that XRPL records ledger close times at 10-second resolution per the protocol; recorded deltas of 1 second do not represent literal 1-second wall-clock intervals.

Methodology refinement note: An earlier version of this article led with per-ledger TPS figures throughout. After publication, Mayukha Vadari (@msvadari) raised the close-time rounding issue publicly. We re-ran the analysis using sustained-window methodology and updated the article. Both methodologies are real and produce real numbers; they measure different things. Thanks to Mayukha for the precision.

References linked in article

XRPLWin is a explorer for XRPL and Xahau. We build tools for the XRPL ecosystem.