diff --git a/akka-distributed-data/src/main/scala/akka/cluster/ddata/ORSet.scala b/akka-distributed-data/src/main/scala/akka/cluster/ddata/ORSet.scala
index 818274b304..f2ac9a6313 100644
--- a/akka-distributed-data/src/main/scala/akka/cluster/ddata/ORSet.scala
+++ b/akka-distributed-data/src/main/scala/akka/cluster/ddata/ORSet.scala
@@ -463,7 +463,8 @@ final class ORSet[A] private[akka] (
         elementsMap
     }
     clearAncestor()
-    val newVvector = vvector.merge(that.vvector)
+
+    val newVvector = vvector.merge(thatDot)
     new ORSet(newElementsMap, newVvector)
   }
 
diff --git a/akka-distributed-data/src/test/scala/akka/cluster/ddata/ORSetSpec.scala b/akka-distributed-data/src/test/scala/akka/cluster/ddata/ORSetSpec.scala
index 32dcaa9f7f..d88bc6c4d6 100644
--- a/akka-distributed-data/src/test/scala/akka/cluster/ddata/ORSetSpec.scala
+++ b/akka-distributed-data/src/test/scala/akka/cluster/ddata/ORSetSpec.scala
@@ -591,6 +591,20 @@ class ORSetSpec extends WordSpec with Matchers {
       b3.merge(a3).merge(c).elements should be(Set.empty)
     }
 
+    "not pollute the vvector of result during mergeRemoveDelta" in {
+      val a = ORSet().add(nodeA, "a")
+      val a1 = a.add(nodeB, "b").remove(nodeB, "b").resetDelta.remove(nodeA, "a") // a1 now contains dot from nodeB
+
+      a.vvector.contains(nodeA) should be(true)
+      a.vvector.contains(nodeB) should be(false)
+      a1.vvector.contains(nodeA) should be(true)
+      a1.vvector.contains(nodeB) should be(true)
+
+      val a2 = ORSet().mergeDelta(a.delta.get).mergeDelta(a1.delta.get)
+      a2.elements should be(Set.empty)
+      a2.vvector.contains(nodeB) should be(false) // a2 should not be polluted by the nodeB dot, as all operations on it pertained only to elements from nodeA
+    }
+
     "have unapply extractor" in {
       val s1 = ORSet.empty.add(node1, "a").add(node2, "b")
       val s2: ORSet[String] = s1