diff --git a/charts/paradedb/docs/runbooks/CNPGClusterLogicalReplicationDistance.md b/charts/paradedb/docs/runbooks/CNPGClusterLogicalReplicationDistance.md
new file mode 100644
index 0000000000..df0e57c885
--- /dev/null
+++ b/charts/paradedb/docs/runbooks/CNPGClusterLogicalReplicationDistance.md
@@ -0,0 +1,34 @@
+# CNPGClusterLogicalReplicationDistance
+
+## Description
+
+These alerts report the subscriber's `received_lsn` minus `latest_end_lsn`.
+Warning fires above 1 GiB for five minutes; critical fires above 4 GiB for two
+minutes. Labels identify the pod, database, and subscription.
+
+## Impact
+
+A large WAL-position gap warrants investigation but does not measure unapplied
+work or committed apply progress. Missing worker metrics cannot trigger these
+alerts; check worker health separately.
+
+## Diagnosis
+
+Inspect the subscription on the subscriber database:
+
+```sql
+SELECT subname, pid, received_lsn, latest_end_lsn,
+       last_msg_receipt_time, latest_end_time,
+       pg_wal_lsn_diff(received_lsn, latest_end_lsn) AS position_gap_bytes
+FROM pg_stat_subscription;
+```
+
+Check receipt age and worker health, and use the
+[logical replication error runbook](CNPGClusterLogicalReplicationErrors.md) to
+inspect apply/sync errors and subscriber logs.
+
+## Mitigation
+
+Address any confirmed connection, worker, or apply errors found during diagnosis.
+Do not skip transactions or resynchronize a subscription based on this position
+gap alone.
diff --git a/charts/paradedb/prometheus_rules/cluster-logical_replication_distance-critical.yaml b/charts/paradedb/prometheus_rules/cluster-logical_replication_distance-critical.yaml
new file mode 100644
index 0000000000..ec5f3adf84
--- /dev/null
+++ b/charts/paradedb/prometheus_rules/cluster-logical_replication_distance-critical.yaml
@@ -0,0 +1,25 @@
+{{- $alert := "CNPGClusterLogicalReplicationDistanceCritical" -}}
+{{- if not (has $alert .excludeRules) -}}
+alert: {{ $alert }}
+annotations:
+  summary: ParadeDB logical replication WAL-position distance is critical
+  description: |-
+    ParadeDB CNPG Cluster "{{ .namespace }}/{{ .cluster }}" reports a WAL-position difference of {{ .value }} GiB as the logical replication subscriber for subscription "{{ .labels.subname }}" in database "{{ .labels.datname }}".
+
+    This does not measure unapplied data. Check replication errors, worker health, and subscriber logs.
+  runbook_url: https://github.com/paradedb/charts/blob/main/charts/paradedb/docs/runbooks/CNPGClusterLogicalReplicationDistance.md
+expr: |
+  max by (namespace, job, pod, datname, subname) (
+    cnpg_pg_stat_subscription_received_lsn{namespace="{{ .namespace }}",pod=~"{{ .podSelector }}"}
+    - cnpg_pg_stat_subscription_latest_end_lsn{namespace="{{ .namespace }}",pod=~"{{ .podSelector }}"}
+  ) / 1024^3 > 4
+for: 2m
+labels:
+  severity: critical
+  namespace: {{ .namespace }}
+  cnpg_cluster: {{ .cluster }}
+  alert_type: logical-replication
+{{- range $key, $val := .additionalLabels }}
+  {{ $key }}: {{ $val | toString | quote }}
+{{- end }}
+{{- end -}}
diff --git a/charts/paradedb/prometheus_rules/cluster-logical_replication_distance-warning.yaml b/charts/paradedb/prometheus_rules/cluster-logical_replication_distance-warning.yaml
new file mode 100644
index 0000000000..004b385244
--- /dev/null
+++ b/charts/paradedb/prometheus_rules/cluster-logical_replication_distance-warning.yaml
@@ -0,0 +1,25 @@
+{{- $alert := "CNPGClusterLogicalReplicationDistanceHigh" -}}
+{{- if not (has $alert .excludeRules) -}}
+alert: {{ $alert }}
+annotations:
+  summary: ParadeDB logical replication WAL-position distance is warning
+  description: |-
+    ParadeDB CNPG Cluster "{{ .namespace }}/{{ .cluster }}" reports a WAL-position difference of {{ .value }} GiB as the logical replication subscriber for subscription "{{ .labels.subname }}" in database "{{ .labels.datname }}".
+
+    This does not measure unapplied data. Check replication errors, worker health, and subscriber logs.
+  runbook_url: https://github.com/paradedb/charts/blob/main/charts/paradedb/docs/runbooks/CNPGClusterLogicalReplicationDistance.md
+expr: |
+  max by (namespace, job, pod, datname, subname) (
+    cnpg_pg_stat_subscription_received_lsn{namespace="{{ .namespace }}",pod=~"{{ .podSelector }}"}
+    - cnpg_pg_stat_subscription_latest_end_lsn{namespace="{{ .namespace }}",pod=~"{{ .podSelector }}"}
+  ) / 1024^3 > 1
+for: 5m
+labels:
+  severity: warning
+  namespace: {{ .namespace }}
+  cnpg_cluster: {{ .cluster }}
+  alert_type: logical-replication
+{{- range $key, $val := .additionalLabels }}
+  {{ $key }}: {{ $val | toString | quote }}
+{{- end }}
+{{- end -}}
diff --git a/charts/paradedb/test/prometheus-rule-rendering/chainsaw-test.yaml b/charts/paradedb/test/prometheus-rule-rendering/chainsaw-test.yaml
index bbf1358283..92d8832b69 100644
--- a/charts/paradedb/test/prometheus-rule-rendering/chainsaw-test.yaml
+++ b/charts/paradedb/test/prometheus-rule-rendering/chainsaw-test.yaml
@@ -78,6 +78,26 @@ spec:
                 exit 1
               fi
 
+              for alert in CNPGClusterLogicalReplicationDistanceHigh CNPGClusterLogicalReplicationDistanceCritical; do
+                distance_rule="$(printf '%s\n' "$rendered" | alert_rule "$alert")"
+                printf '%s\n' "$distance_rule" | grep -Fq 'cnpg_pg_stat_subscription_received_lsn'
+                printf '%s\n' "$distance_rule" | grep -Fq 'cnpg_pg_stat_subscription_latest_end_lsn'
+                printf '%s\n' "$distance_rule" | grep -Fq 'datname, subname'
+                if printf '%s\n' "$distance_rule" | grep -Fq 'kube_customresource'; then
+                  echo "logical distance alert must not require custom-resource metrics"
+                  exit 1
+                fi
+                distance_excluded="$(render 3 --set "cluster.monitoring.prometheusRule.excludeRules[0]=$alert")"
+                if printf '%s\n' "$distance_excluded" | grep -Fq -- "- alert: $alert"; then
+                  echo "excluded logical distance alert must not render"
+                  exit 1
+                fi
+              done
+              printf '%s\n' "$rendered" | alert_rule CNPGClusterLogicalReplicationDistanceHigh | grep -Fq '/ 1024^3 > 1'
+              printf '%s\n' "$rendered" | alert_rule CNPGClusterLogicalReplicationDistanceHigh | grep -Fq 'for: 5m'
+              printf '%s\n' "$rendered" | alert_rule CNPGClusterLogicalReplicationDistanceCritical | grep -Fq '/ 1024^3 > 4'
+              printf '%s\n' "$rendered" | alert_rule CNPGClusterLogicalReplicationDistanceCritical | grep -Fq 'for: 2m'
+
               one_instance="$(render 1)"
               if printf '%s\n' "$one_instance" | grep -Fq 'CNPGClusterHA'; then
                 echo "single-instance clusters must not render HA alerts"