diff --git a/backend/mysql/mysql.go b/backend/mysql/mysql.go
index a8035b79..453d6e44 100644
--- a/backend/mysql/mysql.go
+++ b/backend/mysql/mysql.go
@@ -312,7 +312,8 @@ func (b *mysqlBackend) GetWorkflowInstanceState(ctx context.Context, instance *w
 
 func createInstance(ctx context.Context, tx *sql.Tx, wfi *workflow.Instance, metadata *workflow.Metadata) error {
 	// Check for existing instance
-	if err := tx.QueryRowContext(ctx, "SELECT 1 FROM `instances` WHERE instance_id = ? LIMIT 1", wfi.InstanceID).Scan(new(int)); err != sql.ErrNoRows {
+	if err := tx.QueryRowContext(ctx, "SELECT 1 FROM `instances` WHERE instance_id = ? AND state = ? LIMIT 1", wfi.InstanceID, core.WorkflowInstanceStateActive).
+		Scan(new(int)); err != sql.ErrNoRows {
 		return backend.ErrInstanceAlreadyExists
 	}
 
diff --git a/backend/redis/delete.go b/backend/redis/delete.go
index ba2575b6..31d84f43 100644
--- a/backend/redis/delete.go
+++ b/backend/redis/delete.go
@@ -12,12 +12,9 @@ import (
 // KEYS[2] - pending events key
 // KEYS[3] - history key
 // KEYS[4] - instances-by-creation key
-// KEYS[5] - instances
 // ARGV[1] - instance segment
-// ARGV[2] - instance id
 var deleteCmd = redis.NewScript(
 	`redis.call("DEL", KEYS[1], KEYS[2], KEYS[3])
-	redis.call("HDEL", KEYS[5], ARGV[1])
 	return redis.call("ZREM", KEYS[4], ARGV[1])`)
 
 // deleteInstance deletes an instance from Redis. It does not attempt to remove any future events or pending
@@ -30,8 +27,7 @@ func deleteInstance(ctx context.Context, rdb redis.UniversalClient, instance *co
 		pendingEventsKey(instance),
 		historyKey(instance),
 		instancesByCreation(),
-		instanceIDs(),
-	}, instanceSegment(instance), instance.InstanceID).Err(); err != nil {
+	}, instanceSegment(instance)).Err(); err != nil {
 		return fmt.Errorf("failed to delete instance: %w", err)
 	}
 
diff --git a/backend/redis/expire.go b/backend/redis/expire.go
index 52c8db8e..37dd65fe 100644
--- a/backend/redis/expire.go
+++ b/backend/redis/expire.go
@@ -17,12 +17,10 @@ import (
 // KEYS[3] - instance key
 // KEYS[4] - pending events key
 // KEYS[5] - history key
-// KEYS[6] - instances key
 // ARGV[1] - current timestamp
 // ARGV[2] - expiration time in seconds
 // ARGV[3] - expiration timestamp in unix milliseconds
 // ARGV[4] - instance segment
-// ARGV[5] - instance id
 var expireCmd = redis.NewScript(
 	`-- Find instances which have already expired and remove from the index set
 	local expiredInstances = redis.call("ZRANGE", KEYS[2], "-inf", ARGV[1], "BYSCORE")
@@ -30,7 +28,6 @@ var expireCmd = redis.NewScript(
 		local instanceSegment = expiredInstances[i]
 		redis.call("ZREM", KEYS[1], instanceSegment) -- index set
 		redis.call("ZREM", KEYS[2], instanceSegment) -- expiration set
-		redis.call("HDEL", KEYS[6], ARGV[5])
 	end
 
 	-- Add expiration time for future cleanup
@@ -58,12 +55,10 @@ func setWorkflowInstanceExpiration(ctx context.Context, rdb redis.UniversalClien
 		instanceKey(instance),
 		pendingEventsKey(instance),
 		historyKey(instance),
-		instanceIDs(),
 	},
 		nowStr,
 		expiration.Seconds(),
 		expStr,
 		instanceSegment(instance),
-		instance.InstanceID,
 	).Err()
 }
diff --git a/backend/redis/instance.go b/backend/redis/instance.go
index 685ef156..3526aab5 100644
--- a/backend/redis/instance.go
+++ b/backend/redis/instance.go
@@ -165,9 +165,6 @@ func createInstanceP(ctx context.Context, p redis.Pipeliner, instance *core.Work
 	// The newly created instance is going to be the active execution
 	setActiveInstanceExecutionP(ctx, p, instance)
 
-	// Record instance id
-	p.HSet(ctx, instanceIDs(), instance.InstanceID, 1)
-
 	p.ZAdd(ctx, instancesByCreation(), redis.Z{
 		Member: instanceSegment(instance),
 		Score:  float64(createdAt.UnixMilli()),
diff --git a/backend/redis/keys.go b/backend/redis/keys.go
index 8ebfa873..24f6d2bf 100644
--- a/backend/redis/keys.go
+++ b/backend/redis/keys.go
@@ -37,10 +37,6 @@ func instancesExpiring() string {
 	return "instances-expiring"
 }
 
-func instanceIDs() string {
-	return "instances"
-}
-
 func pendingEventsKey(instance *core.WorkflowInstance) string {
 	return fmt.Sprintf("pending-events:%v", instanceSegment(instance))
 }
diff --git a/backend/redis/scripts/complete_workflow_task.lua b/backend/redis/scripts/complete_workflow_task.lua
index 36e3bb4c..93de72aa 100644
--- a/backend/redis/scripts/complete_workflow_task.lua
+++ b/backend/redis/scripts/complete_workflow_task.lua
@@ -21,7 +21,6 @@ local pendingEventsKey = getKey()
 local payloadHashKey = getKey()
 local futureEventZSetKey = getKey()
 local activeInstancesKey = getKey()
-local instancesByIdKey = getKey()
 
 local workflowSetKey = getKey()
 local workflowStreamKey = getKey()
@@ -103,7 +102,7 @@ for i = 1, otherWorkflowInstances do
         local conflictEventPayloadData = getArgv()
 
         -- Does the instance exist already?
-        local instanceExists = redis.call("HEXISTS", instancesByIdKey, targetInstanceId)
+        local instanceExists = redis.call("EXISTS", targetActiveInstanceExecutionState)
         if instanceExists == 1 then
             redis.call("XADD", pendingEventsKey, "*", "event", conflictEventData)
             storePayload(conflictEventId, conflictEventPayloadData)
@@ -119,8 +118,6 @@ for i = 1, otherWorkflowInstances do
 
             -- Track active instance
             redis.call("SADD", activeInstancesKey, targetInstanceSegment)
-
-            redis.call("HSET", instancesByIdKey, targetInstanceId, 1)
         end
     end
 
diff --git a/backend/redis/workflow.go b/backend/redis/workflow.go
index 402e5e13..fb68fa95 100644
--- a/backend/redis/workflow.go
+++ b/backend/redis/workflow.go
@@ -106,7 +106,6 @@ func (rb *redisBackend) CompleteWorkflowTask(
 		payloadKey(instance),
 		futureEventsKey(),
 		instancesActive(),
-		instanceIDs(),
 		queueKeys.SetKey,
 		queueKeys.StreamKey,
 	)
diff --git a/backend/sqlite/sqlite.go b/backend/sqlite/sqlite.go
index 791f02c6..35c06174 100644
--- a/backend/sqlite/sqlite.go
+++ b/backend/sqlite/sqlite.go
@@ -187,7 +187,8 @@ func (sb *sqliteBackend) CreateWorkflowInstance(ctx context.Context, instance *w
 
 func createInstance(ctx context.Context, tx *sql.Tx, wfi *workflow.Instance, metadata *workflow.Metadata) error {
 	// Check for existing instance
-	if err := tx.QueryRowContext(ctx, "SELECT 1 FROM `instances` WHERE id = ? LIMIT 1", wfi.InstanceID).Scan(new(int)); err != sql.ErrNoRows {
+	if err := tx.QueryRowContext(ctx, "SELECT 1 FROM `instances` WHERE id = ? AND state = ? LIMIT 1", wfi.InstanceID, core.WorkflowInstanceStateActive).
+		Scan(new(int)); err != sql.ErrNoRows {
 		return backend.ErrInstanceAlreadyExists
 	}
 
diff --git a/backend/test/e2e.go b/backend/test/e2e.go
index fd1a3490..9794681e 100644
--- a/backend/test/e2e.go
+++ b/backend/test/e2e.go
@@ -206,18 +206,25 @@ func EndToEndBackendTest(t *testing.T, setup func(options ...backend.BackendOpti
 			name: "SubWorkflow_DuplicateInstanceID",
 			f: func(t *testing.T, ctx context.Context, c *client.Client, w *worker.Worker, b TestBackend) {
 				swf := func(ctx workflow.Context, i int) (int, error) {
+					workflow.NewSignalChannel[any](ctx, "signal").Receive(ctx)
+
 					return i * 2, nil
 				}
 				wf := func(ctx workflow.Context) (int, error) {
-					r, err := workflow.CreateSubWorkflowInstance[int](ctx, workflow.SubWorkflowOptions{
+					swf1 := workflow.CreateSubWorkflowInstance[int](ctx, workflow.SubWorkflowOptions{
 						InstanceID: "subworkflow",
-					}, swf, 1).Get(ctx)
-					if err != nil {
-						return 0, err
-					}
+					}, swf, 1)
+
+					defer func() {
+						rctx := workflow.NewDisconnectedContext(ctx)
+
+						// Unblock waiting sub workflow
+						workflow.SignalWorkflow[any](rctx, "subworkflow", "signal", 1).Get(rctx)
+						swf1.Get(rctx)
+					}()
 
 					// Run another subworkflow with the same ID
-					r, err = workflow.CreateSubWorkflowInstance[int](ctx, workflow.SubWorkflowOptions{
+					r, err := workflow.CreateSubWorkflowInstance[int](ctx, workflow.SubWorkflowOptions{
 						InstanceID: "subworkflow",
 					}, swf, 1).Get(ctx)
 					if err != nil {