A test pass answers a defined laboratory question
A WaterSense toilet performance pass shows that a specified toilet configuration met applicable test criteria. It does not give a household's probability of needing a second flush. Read the test procedure, tested configuration and specification version together; a gallons-per-flush number alone cannot answer a bowl-clearing question.
This guide explains how to interpret that evidence when comparing tank-type toilets. It contains an original decision table and a small test-outcome example, not a product test, brand ranking or installation diagnosis.
Check the effective specification before using a newer PDF
At the November 8, 2026 source check, EPA's residential toilet page said Version 1.2 remains in effect and the effective date of Version 2.0 is paused. A published document and an effective requirement are different records. Recheck that status before making a current certification claim.
Version 1.2, section 4.1, describes complete clearance of test media in one flush in at least four of five attempts, using the referenced waste-extraction protocol. EPA's performance overview explains the 350-gram soybean-paste load and four balls of toilet paper behind the criterion. This is controlled test media, not a claim that household waste is uniform.
Keep different performance questions separate
EPA describes several tests on its residential toilet page. Use the following interpretation worksheet to avoid treating one result as an answer to every concern. The right-hand column is our evidence check, not an additional EPA requirement.
| Evidence category | Question it addresses | What still needs separate evidence |
|---|---|---|
| Waste extraction | Clearing specified media from the bowl | How the exact installed fixture behaves in your setting |
| Bowl surface cleaning | Cleaning the bowl surface under the test | Your cleaning routine and observations |
| Drainline transport | Moving test media through a test drainline | The condition and arrangement of your actual drain |
| Mixed-media flushability | Handling defined media of differing properties | Any claim about an untested object or operating condition |
For example, a comparison sheet that supplies only a waste-extraction result leaves the other questions unanswered. Write “not supplied” rather than copying that result into every column. Conversely, an absent number on a retailer's page does not prove a product failed a test.
Why four of five does not mean an 80% household success rate
Consider two invented five-attempt records under the same defined test: A is clear, clear, clear, clear, not clear; B is clear, clear, clear, not clear, not clear. A has four clearances and B has three. Their arithmetic fractions are 4 ÷ 5 = 80% and 3 ÷ 5 = 60%. A meets the stated count criterion; B does not. These records do not establish certification by themselves.
The fractions describe those five attempts. They are not predictions that A will fail one in five household uses or that B will use a particular annual water volume. A household forecast would require a relevant observation period, definitions for a use and a repeat flush, and data from the setting being forecast. The laboratory threshold supplies none of those household counts.
A second trap is turning a pass into a ranking. If two documented configurations both satisfy the criterion, the pass statements alone do not say which is quieter, more comfortable or more suitable for a particular drain. Those are additional decisions requiring their own evidence.
Build a comparison record you can actually check
Record the full model identity, tank and bowl combination where applicable, source URL, document date, effective specification, and the exact claim being compared. Keep test evidence separate from marketing language. For an existing installation problem, document the symptom before attributing it to a nominal flush rating; this article has not inspected a fixture or drain.
Use the toilet research directory for historical rating context and the rough-in and footprint checklist for physical fit. Neither a fitting dimension nor a saved directory record substitutes for the current, exact configuration's certification evidence. The useful result is a comparison with explicit unknowns, rather than a single number asked to stand for every kind of performance.