Marfa
What Marfa filed with TWDB about the water it produced and the water it billed for, 2015–2024 — and what the filing does and does not let anyone verify.
- Non-revenue water, FY2024
- 44.8% Computed from filed system input and billed consumption.
- system input volume 181.1 sheet 2024, cell AP282 · converted to million gallons from the unit the filing states
- billed authorized consumption 100 sheet 2024, cell AU282 · converted to million gallons from the unit the filing states
- Annual value of the filed loss
- $68,600 Apparent loss is valued at this utility's own reported retail price of water, and real loss at its own reported variable production cost. Not our estimate, and shown to three significant figures.
- cost of apparent losses, as filed 39,423 sheet 2024, cell CD282 · republished only where the retail price it implies is one a utility could charge
- cost of real losses, as filed 29,192 sheet 2024, cell CH282 · republished only where the production cost it implies is one a utility could incur
- Apparent loss that is a default
- 8.7%
- apparent loss 5.76 sheet 2024, cell BL282 · converted to million gallons from the unit the filing states
2026-08-11-7270fa7072e5.xlsx2024 AP282agency source
Where this figure comes from
(system input − billed authorized) ÷ system input × 100
Read from data/archive/twdb/2026-08-11-7270fa7072e5.xlsx (SHA-256 7270fa7072e51e70), our archived copy of the file as it stood on 2026-08-11. That path is committed to the repository and is not served here. The agency publishes its own current copy at https://www.twdb.texas.gov/conservation/data/doc/water-loss-data/sources/historical-water-loss-audit-data.xlsx, which may since have been revised.
2026-08-11-7270fa7072e5.xlsx2024 CD282agency source
Where this figure comes from
cost of apparent losses + cost of real losses, both as computed in the filed audit
Read from data/archive/twdb/2026-08-11-7270fa7072e5.xlsx (SHA-256 7270fa7072e51e70), our archived copy of the file as it stood on 2026-08-11. That path is committed to the repository and is not served here. The agency publishes its own current copy at https://www.twdb.texas.gov/conservation/data/doc/water-loss-data/sources/historical-water-loss-audit-data.xlsx, which may since have been revised.
MeasuredTWDB default
2026-08-11-7270fa7072e5.xlsx2024 BL282agency source
Where this figure comes from
defaulted apparent-loss components ÷ total apparent loss × 100
Read from data/archive/twdb/2026-08-11-7270fa7072e5.xlsx (SHA-256 7270fa7072e51e70), our archived copy of the file as it stood on 2026-08-11. That path is committed to the repository and is not served here. The agency publishes its own current copy at https://www.twdb.texas.gov/conservation/data/doc/water-loss-data/sources/historical-water-loss-audit-data.xlsx, which may since have been revised.
What the filing says
Marfa serves 1,639 retail connections and about 2,500 people. Because it reports fewer than 3,300 connections, it is on a five-year water loss audit cycle with TWDB.
What the record shows
44.8% non-revenue water in FY2024, higher than 91% of the 908 Texas systems that filed that year.
- The figure is getting worse across the record.
Compliance record
8 violations on EPA's record for this system as of 2026Q2, of which 3 are health-based. 4 resolved, 4 archived, and 0 still open.
Nothing on the record is open as of 2026Q2.
Counted as distinct violations, not as rows: EPA's export carries one row per enforcement action taken against a violation, so a row count would multiply this system's record several times over. A violation here is a breach of a reporting, monitoring or contaminant-level rule as reported to EPA — it is not a measurement of the water, and this site has none.
Source: US EPA, Safe Drinking Water Information System, via the ECHO national download, 2026Q2. How we hold it.
- FY2019 reports a leakage index of 0.06, which would mean less real loss than the unavoidable minimum the same audit computes for this network.
- FY2018 reports a leakage index of 0.07, which would mean less real loss than the unavoidable minimum the same audit computes for this network.
Each is reproduced exactly as filed and excluded from every figure computed on this page. A number that contradicts its own audit is a finding about the filing, not a gap in ours.
TWDB derives water losses from system input minus authorized consumption, so these parts sum to the whole on every filing. What a filing can get wrong is which of these numbers were measured.
Where these volumes come from
- System input 181.1 MG sheet 2024, cell AP282 · converted to million gallons from the unit the filing states
- Billed authorized 100 MG sheet 2024, cell AU282 · converted to million gallons from the unit the filing states
- Unbilled authorized 70.3 MG computed here, not read from a cell
- Apparent loss 5.76 MG sheet 2024, cell BL282 · converted to million gallons from the unit the filing states
- Real loss 5.03 MG sheet 2024, cell BQ282
From data/archive/twdb/2026-08-11-7270fa7072e5.xlsx (SHA-256 7270fa7072e51e70), retrieved 2026-08-11 from the published source.
| Year | NRW | Input (MG) | Recoverable |
|---|---|---|---|
| 2024 | 44.8% | 181.1 | $68,600 |
| 2023 | 40.9% | 181.3 | $80,800 |
| 2022 | 49.6% | 176.8 | $66,600 |
| 2021 | 41.0% | 171.4 | $24,200 |
| 2020 | 42.9% | 188.7 | $30,200 |
| 2019 | 39.2% | 176 | $27,700 |
| 2018 | 38.0% | 172.6 | $27,800 |
| 2017 | 43.5% | 179.1 | $32,800 |
| 2016 | 41.7% | 166.1 | $35,400 |
| 2015 | 38.0% | 160.5 | $30,600 |
Balance closure is inapplicable on every filed year above, so the column that repeated it is not shown. TWDB derives water losses from system input minus authorized consumption, so this decomposition holds by construction. This is neither a pass nor a disclosure gap.
How much of this figure was measured
Apparent loss is three components: customer meter inaccuracy, unauthorized consumption, and data handling error. TWDB supplies a default value for the last two when a utility does not measure them, and records which in its export. Across the published record, unauthorized consumption is a default on 95% of filings, data handling error on 30%. The median filing is 14.4% defaulted, and 1,329 filings are more than half. TWDB's own data-validity score averages 68.4 out of 100 on the 9,315 filings using a default, against 70.0 on the 429 using none — so the score does not tell the two apart.
8.7% of Marfa's reported apparent loss for FY2024 is a default rather than a measurement. How much of a Texas loss figure is measured sets this out across the whole filed record.
How the loss is valued
Apparent loss is water that reached a customer and was never billed: under-registering meters, data-handling errors, unauthorized use. It is valued here at the retail rate, because water that reached a customer would have sold at that rate. Real loss is valued at variable production cost only, since eliminating a leak saves what it cost to treat and pump the water rather than what it would have sold for. Valuing both at the retail rate is how a loss study produces a number nobody in a finance department believes, and that is the mistake this figure is built to avoid.
This is the value of the loss, not an amount anyone has established is recoverable. How much of it a utility can actually collect depends on which meters are under-registering, which accounts are unbilled, and what a replacement or an audit of those accounts would cost. None of that is in a filed water loss audit, and this page does not claim to know it.
Apparent loss is valued at this utility's own reported retail price of water, and real loss at its own reported variable production cost. Not our estimate, and shown to three significant figures.
Comparable systems
In the same county.
| System | Basis for comparison | NRW |
|---|---|---|
| Presidio | PRESIDIO County | 24.3% FY2015 |
| Candelaria Water Supply Corp | PRESIDIO County | 41.0% FY2022 |
| Redford Water Supply | PRESIDIO County | 13.7% FY2023 |
| Brazoria County MUD 21 | 1,000–3,300 connections | 14.5% FY2021 |
| Harris County MUD 276 | 1,000–3,300 connections | 18.3% FY2020 |
| Harris County MUD 64 | 1,000–3,300 connections | 6.0% FY2020 |
| Henrietta | 1,000–3,300 connections | 13.2% FY2024 |
| West Harris County MUD 2 Chase | 1,000–3,300 connections | 4.6% FY2020 |
Where is it going?
This page values the loss at roughly $68,600 a year. It cannot say which accounts, which zones, or which meters, and it cannot say how much of that value is collectable. Answering either takes your billing history, your meter reads, and your production data reconciled against each other.
Our Water Loss Recovery Snapshot is three weeks, one flat fee, and the deliverable is a ranked list of specific accounts and zones with dollars and a confidence on each.
See what the Snapshot delivers Email us about MarfaSources and corrections. Every figure above is either filed by the utility or computed from filed figures, and each one says which. If a number here looks wrong, tell us: the error may be in the filing, in our reading of it, or in something the export does not show. Send a correction. We will check it against the source, correct the page where it is wrong, and say what changed.
- Texas Water Development Board — historical water loss audit data. Retrieved 2026-08-11 from data/archive/twdb/2026-08-11-7270fa7072e5.xlsx. https://www.twdb.texas.gov/conservation/data/doc/water-loss-data/sources/historical-water-loss-audit-data.xlsx