diff --git a/CHANGELOG.md b/CHANGELOG.md
index 263d126d..9abc27b4 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -14,6 +14,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Fixed
 
 - Fixed periodic jobs advancing their durable next run time when job insertion fails. [PR #1359](https://github.com/riverqueue/river/pull/1359).
+- Improved SQLite queue count performance on large job tables by limiting counts to available and running jobs so the existing state and queue index can be used. [PR #1360](https://github.com/riverqueue/river/pull/1360).
 
 ## [0.44.1] - 2026-08-21
 
diff --git a/riverdriver/riverdrivertest/job_read.go b/riverdriver/riverdrivertest/job_read.go
index fd6ad6d1..f8fe4418 100644
--- a/riverdriver/riverdrivertest/job_read.go
+++ b/riverdriver/riverdrivertest/job_read.go
@@ -116,6 +116,34 @@ func exerciseJobRead[TTx any](ctx context.Context, t *testing.T, executorWithTx
 			require.Equal(t, int64(1), countsByQueue[1].CountRunning)
 		})
 
+		t.Run("IgnoresJobsInOtherStates", func(t *testing.T) {
+			t.Parallel()
+
+			exec, _ := setup(ctx, t)
+
+			for _, state := range []rivertype.JobState{
+				rivertype.JobStateCancelled,
+				rivertype.JobStateCompleted,
+				rivertype.JobStateDiscarded,
+				rivertype.JobStatePending,
+				rivertype.JobStateRetryable,
+				rivertype.JobStateScheduled,
+			} {
+				_ = testfactory.Job(ctx, t, exec, &testfactory.JobOpts{Queue: new("queue1"), State: new(state)})
+			}
+
+			countsByQueue, err := exec.JobCountByQueueAndState(ctx, &riverdriver.JobCountByQueueAndStateParams{
+				QueueNames: []string{"queue1"},
+				Schema:     "",
+			})
+			require.NoError(t, err)
+
+			require.Len(t, countsByQueue, 1)
+			require.Equal(t, "queue1", countsByQueue[0].Queue)
+			require.Equal(t, int64(0), countsByQueue[0].CountAvailable)
+			require.Equal(t, int64(0), countsByQueue[0].CountRunning)
+		})
+
 		t.Run("IncludesRequestedQueuesThatHaveNoJobs", func(t *testing.T) {
 			t.Parallel()
 
diff --git a/riverdriver/riversqlite/internal/dbsqlc/river_job.sql b/riverdriver/riversqlite/internal/dbsqlc/river_job.sql
index 6f8bd9a4..dd284e7e 100644
--- a/riverdriver/riversqlite/internal/dbsqlc/river_job.sql
+++ b/riverdriver/riversqlite/internal/dbsqlc/river_job.sql
@@ -63,6 +63,7 @@ WITH queue_stats AS (
         COUNT(CASE WHEN river_job.state = 'running' THEN 1 END) AS count_running
     FROM /* TEMPLATE: schema */river_job
     WHERE river_job.queue IN (sqlc.slice('queue_names'))
+        AND river_job.state IN ('available', 'running')
     GROUP BY river_job.queue
 )
 
diff --git a/riverdriver/riversqlite/internal/dbsqlc/river_job.sql.go b/riverdriver/riversqlite/internal/dbsqlc/river_job.sql.go
index 61ef76b9..68701243 100644
--- a/riverdriver/riversqlite/internal/dbsqlc/river_job.sql.go
+++ b/riverdriver/riversqlite/internal/dbsqlc/river_job.sql.go
@@ -109,6 +109,7 @@ WITH queue_stats AS (
         COUNT(CASE WHEN river_job.state = 'running' THEN 1 END) AS count_running
     FROM /* TEMPLATE: schema */river_job
     WHERE river_job.queue IN (/*SLICE:queue_names*/?)
+        AND river_job.state IN ('available', 'running')
     GROUP BY river_job.queue
 )