diff --git a/src/main/scala/com/yahoo/bullet/spark/JoinStreaming.scala b/src/main/scala/com/yahoo/bullet/spark/JoinStreaming.scala
index e0bc8ed..e4e7be4 100644
--- a/src/main/scala/com/yahoo/bullet/spark/JoinStreaming.scala
+++ b/src/main/scala/com/yahoo/bullet/spark/JoinStreaming.scala
@@ -122,7 +122,7 @@ object JoinStreaming {
     }
 
     // If the query is not finished and data is not empty, consume data and check if it should be emitted.
-    if (!querierState.finished && filterResultData.data.nonEmpty) {
+    if (!querierState.finished && filterResultData.data != null && filterResultData.data.nonEmpty) {
       querierState.querier.combine(filterResultData.data)
       emitResults(key, querierState, bulletResults, metrics)
     }
diff --git a/src/test/scala/com/yahoo/bullet/spark/JoinStreamingTest.scala b/src/test/scala/com/yahoo/bullet/spark/JoinStreamingTest.scala
index 2ceb253..bc88453 100644
--- a/src/test/scala/com/yahoo/bullet/spark/JoinStreamingTest.scala
+++ b/src/test/scala/com/yahoo/bullet/spark/JoinStreamingTest.scala
@@ -73,6 +73,36 @@ class JoinStreamingTest extends BulletSparkTest {
     }
   }
 
+  it should "handle queries for which there is no state" in {
+    val inputQueries: mutable.Queue[RDD[(String, BulletData)]] = mutable.Queue()
+    val outputCollector = ListBuffer.empty[Array[(String, BulletResult)]]
+
+    val pubSubMessage = new PubSubMessage("id",
+      makeSimpleAggregationFilterQuery("field", List("b235gf23b").asJava, Operation.EQUALS, RAW, 1))
+    val runningQuery = new RunningQuery("id", pubSubMessage.getContent, config)
+
+    val inputQueryStream = ssc.queueStream(inputQueries)
+
+    val broadcastedConfig = BulletSparkConfig.getInstance(ssc, config)
+    val outputStream = JoinStreaming.join(ssc, inputQueryStream, broadcastedConfig)
+
+    outputStream.foreachRDD(rdd => outputCollector += rdd.collect())
+
+    ssc.checkpoint("target/spark-test")
+
+    ssc.start()
+
+    // Consuming a RunningQuery in the join stage for which there is no state yet
+    val bulletQuery = BulletSparkUtils.createBulletQuerier(runningQuery, Querier.Mode.PARTITION, broadcastedConfig)
+    inputQueries += sc.makeRDD(Seq(("id", new FilterResultData(metadata, runningQuery, bulletQuery.getData))))
+
+    wait1second() // T = 1s
+
+    eventually {
+      outputCollector.last.length should equal(0)
+    }
+  }
+
   it should "output join results on query timeout" in {
     val inputQueries: mutable.Queue[RDD[(String, BulletData)]] = mutable.Queue()
     val outputCollector = ListBuffer.empty[Array[(String, BulletResult)]]