diff --git a/datasette/facets.py b/datasette/facets.py
index ccd85461..f49575d9 100644
--- a/datasette/facets.py
+++ b/datasette/facets.py
@@ -65,6 +65,8 @@ def register_facet_classes():
 
 class Facet:
     type = None
+    # How many rows to consider when suggesting facets:
+    suggest_consider = 1000
 
     def __init__(
         self,
@@ -145,17 +147,6 @@ class Facet:
             )
         ).columns
 
-    async def get_row_count(self):
-        if self.row_count is None:
-            self.row_count = (
-                await self.ds.execute(
-                    self.database,
-                    f"select count(*) from ({self.sql})",
-                    self.params,
-                )
-            ).rows[0][0]
-        return self.row_count
-
 
 class ColumnFacet(Facet):
     type = "column"
@@ -170,13 +161,16 @@ class ColumnFacet(Facet):
             if column in already_enabled:
                 continue
             suggested_facet_sql = """
-                select {column} as value, count(*) as n from (
-                    {sql}
-                ) where value is not null
+                with limited as (select * from ({sql}) limit {suggest_consider})
+                select {column} as value, count(*) as n from limited
+                where value is not null
                 group by value
                 limit {limit}
             """.format(
-                column=escape_sqlite(column), sql=self.sql, limit=facet_size + 1
+                column=escape_sqlite(column),
+                sql=self.sql,
+                limit=facet_size + 1,
+                suggest_consider=self.suggest_consider,
             )
             distinct_values = None
             try:
@@ -211,6 +205,17 @@ class ColumnFacet(Facet):
                 continue
         return suggested_facets
 
+    async def get_row_count(self):
+        if self.row_count is None:
+            self.row_count = (
+                await self.ds.execute(
+                    self.database,
+                    f"select count(*) from (select * from ({self.sql}) limit {self.suggest_consider})",
+                    self.params,
+                )
+            ).rows[0][0]
+        return self.row_count
+
     async def facet_results(self):
         facet_results = []
         facets_timed_out = []
@@ -313,11 +318,14 @@ class ArrayFacet(Facet):
                 continue
             # Is every value in this column either null or a JSON array?
             suggested_facet_sql = """
+                with limited as (select * from ({sql}) limit {suggest_consider})
                 select distinct json_type({column})
-                from ({sql})
+                from limited
                 where {column} is not null and {column} != ''
             """.format(
-                column=escape_sqlite(column), sql=self.sql
+                column=escape_sqlite(column),
+                sql=self.sql,
+                suggest_consider=self.suggest_consider,
             )
             try:
                 results = await self.ds.execute(
@@ -402,7 +410,9 @@ class ArrayFacet(Facet):
                 order by
                     count(*) desc, value limit {limit}
             """.format(
-                col=escape_sqlite(column), sql=self.sql, limit=facet_size + 1
+                col=escape_sqlite(column),
+                sql=self.sql,
+                limit=facet_size + 1,
             )
             try:
                 facet_rows_results = await self.ds.execute(
diff --git a/tests/test_facets.py b/tests/test_facets.py
index 023efcf0..a2b505ec 100644
--- a/tests/test_facets.py
+++ b/tests/test_facets.py
@@ -1,6 +1,6 @@
 from datasette.app import Datasette
 from datasette.database import Database
-from datasette.facets import ColumnFacet, ArrayFacet, DateFacet
+from datasette.facets import Facet, ColumnFacet, ArrayFacet, DateFacet
 from datasette.utils.asgi import Request
 from datasette.utils import detect_json1
 from .fixtures import make_app_client
@@ -662,3 +662,39 @@ async def test_facet_against_in_memory_database():
     assert response1.status_code == 200
     response2 = await ds.client.get("/mem/t?_facet=name&_facet=name2")
     assert response2.status_code == 200
+
+
+@pytest.mark.asyncio
+async def test_facet_only_considers_first_x_rows():
+    # This test works by manually fiddling with Facet.suggest_consider
+    ds = Datasette()
+    original_suggest_consider = Facet.suggest_consider
+    try:
+        Facet.suggest_consider = 40
+        db = ds.add_memory_database("test_facet_only_x_rows")
+        await db.execute_write("create table t (id integer primary key, col text)")
+        # First 50 rows make it look like col and col_json should be faceted
+        to_insert = [{"col": "one" if i % 2 else "two"} for i in range(50)]
+        await db.execute_write_many("insert into t (col) values (:col)", to_insert)
+        # Next 50 break that assumption
+        to_insert2 = [{"col": f"x{i}"} for i in range(50)]
+        await db.execute_write_many("insert into t (col) values (:col)", to_insert2)
+        response = await ds.client.get(
+            "/test_facet_only_x_rows/t.json?_extra=suggested_facets"
+        )
+        data = response.json()
+        assert data["suggested_facets"] == [
+            {
+                "name": "col",
+                "toggle_url": "http://localhost/test_facet_only_x_rows/t.json?_extra=suggested_facets&_facet=col",
+            }
+        ]
+        # But if we set suggest_consider to 100 they are not suggested
+        Facet.suggest_consider = 100
+        response2 = await ds.client.get(
+            "/test_facet_only_x_rows/t.json?_extra=suggested_facets"
+        )
+        data2 = response2.json()
+        assert data2["suggested_facets"] == []
+    finally:
+        Facet.suggest_consider = original_suggest_consider