diff --git a/tests/contracts/README.md b/tests/contracts/README.md
index 3599c99c..c8d24e1d 100644
--- a/tests/contracts/README.md
+++ b/tests/contracts/README.md
@@ -9,8 +9,9 @@ The suite uses four dependency-oriented layers without moving established tests:
   `wqp_test.py`, `nldi_test.py`, `streamstats_test.py`): service request construction,
   response parsing, and documented protocol behavior.
 - **Component** (`transport_test.py`, `waterdata_chunking_test.py`,
-  `waterdata_queryables_test.py`, `waterdata_endpoints_test.py`, `rdb_test.py`,
-  `_csv_test.py`): one internal responsibility in isolation.
+  `waterdata_queryables_test.py`, `waterdata_endpoints_test.py`,
+  `waterdata_properties_test.py`, `rdb_test.py`, `_csv_test.py`): one internal
+  responsibility in isolation.
 - **Cross-component** (`architecture_test.py`, `headers_host_scoping_test.py`,
   `waterdata_progress_test.py`): dependency fitness functions and behavior that
   spans adapters, OGC, transport, or security boundaries.
diff --git a/tests/waterdata_properties_test.py b/tests/waterdata_properties_test.py
new file mode 100644
index 00000000..cb8dc0e5
--- /dev/null
+++ b/tests/waterdata_properties_test.py
@@ -0,0 +1,72 @@
+"""Live monitor: the columns a getter documents are the ones its collection has.
+
+Some getters list their returned columns in the ``properties`` docstring, under
+"Available options are:". The list is hand-written, so it goes stale when USGS
+adds or removes a field. It is checked here rather than generated, because a
+generated list would make the docs build depend on the live service and would
+not appear in ``help()``.
+"""
+
+import inspect
+import re
+
+import pytest
+
+from dataretrieval import waterdata
+from dataretrieval.ogc.schema import _check_ogc_requests
+from dataretrieval.waterdata.endpoints import ogc_api_url
+
+#: Getters that list their returned columns, by collection. Written out rather
+#: than discovered, so a getter that loses its list fails instead of being
+#: skipped. get_channel and get_monitoring_locations also list columns but are
+#: left out until their lists match the schema.
+_DOCUMENTED = {
+    "daily": waterdata.get_daily,
+    "continuous": waterdata.get_continuous,
+    "latest-continuous": waterdata.get_latest_continuous,
+    "latest-daily": waterdata.get_latest_daily,
+    "time-series-metadata": waterdata.get_time_series_metadata,
+}
+
+#: The label may be split across two lines. The list ends at the next unindented
+#: line of the dedented docstring, which is the next numpydoc parameter.
+_COLUMNS_RE = re.compile(r"Available\s+options\s+are:(.*?)(?=\n\S|\Z)", re.S)
+
+#: Some schemas list ``id`` and some do not, but every getter accepts it, so it
+#: is excluded from both sides.
+_ALWAYS_REQUESTABLE = {"id"}
+
+
+def _documented_properties(getter) -> set[str]:
+    """The column names *getter* lists in its ``properties`` docstring."""
+    match = _COLUMNS_RE.search(inspect.getdoc(getter) or "")
+    assert match, f"{getter.__name__} no longer documents its columns"
+    return {n.strip().rstrip(".") for n in match.group(1).split(",") if n.strip()}
+
+
+def _schema_properties(collection: str) -> set[str]:
+    """The columns *collection* publishes in its OGC schema document."""
+    body, _ = _check_ogc_requests(collection, "schema", base_url=ogc_api_url())
+    properties = body.get("properties")
+    assert properties, f"{collection} published no schema properties"
+    return set(properties)
+
+
+@pytest.mark.live
+@pytest.mark.parametrize("collection", sorted(_DOCUMENTED))
+def test_documented_columns_match_the_collection_schema(collection):
+    """A getter's documented column list matches what the collection publishes.
+
+    For an added field, consider a named parameter too; a removed field is a
+    breaking change and belongs in NEWS.
+    """
+    getter = _DOCUMENTED[collection]
+    documented = _documented_properties(getter) - _ALWAYS_REQUESTABLE
+    published = _schema_properties(collection) - _ALWAYS_REQUESTABLE
+
+    assert documented == published, (
+        f"{getter.__name__} documents the wrong columns for {collection}: "
+        f"missing={sorted(published - documented)}, "
+        f"stale={sorted(documented - published)}. Edit the 'Available options "
+        "are:' list in its properties docstring."
+    )