diff --git a/src/main/java/com/teragrep/pth10/steps/teragrep/TeragrepBloomStep.java b/src/main/java/com/teragrep/pth10/steps/teragrep/TeragrepBloomStep.java
index e5f34574..d20ae856 100644
--- a/src/main/java/com/teragrep/pth10/steps/teragrep/TeragrepBloomStep.java
+++ b/src/main/java/com/teragrep/pth10/steps/teragrep/TeragrepBloomStep.java
@@ -157,7 +157,7 @@ private Dataset<Row> updateBloomFilter(Dataset<Row> dataset) {
 
     private Dataset<Row> estimateSize(Dataset<Row> dataset) {
         return dataset
-                .select(functions.col("partition"), functions.explode(functions.col(inputCol)).as("token"))
+                .select(functions.col("partition"), functions.explode_outer(functions.col(inputCol)).as("token"))
                 .groupBy("partition")
                 .agg(functions.approxCountDistinct("token").as(outputCol));
     }
diff --git a/src/test/java/com/teragrep/pth10/BloomFilterOperationsTest.java b/src/test/java/com/teragrep/pth10/BloomFilterOperationsTest.java
index b2e5321c..0dc53c5b 100644
--- a/src/test/java/com/teragrep/pth10/BloomFilterOperationsTest.java
+++ b/src/test/java/com/teragrep/pth10/BloomFilterOperationsTest.java
@@ -128,6 +128,34 @@ public void estimateTest() {
                 );
     }
 
+    @Test
+    @DisabledIfSystemProperty(
+            named = "skipSparkTest",
+            matches = "true"
+    )
+    public void testEstimateOnEmptyArray() {
+        streamingTestUtil
+                .performDPLTest(
+                        // index_Empty _raw = "" so tokenizer step will produce an empty array
+                        "index=index_Empty earliest=2020-01-01T00:00:00z latest=2023-01-01T00:00:00z | teragrep exec tokenizer | teragrep exec bloom estimate",
+                        testFile, ds -> {
+                            Assertions
+                                    .assertEquals("[partition, estimate(tokens)]", Arrays.toString(ds.columns()), "Batch handler dataset contained an unexpected column arrangement !");
+                            List<Integer> results = ds
+                                    .select("estimate(tokens)")
+                                    .collectAsList()
+                                    .stream()
+                                    .map(r -> Integer.parseInt(r.get(0).toString()))
+                                    .collect(Collectors.toList());
+
+                            // assert that a row is produced and not an empty dataframe
+                            Assertions.assertEquals(1, results.size());
+                            // assert that estimate is 0 and not empty or null
+                            Assertions.assertEquals(0, results.get(0));
+                        }
+                );
+    }
+
     @Test
     @DisabledIfSystemProperty(
             named = "skipSparkTest",
diff --git a/src/test/resources/xmlWalkerTestDataStreaming/bloomTeragrepStep_data.jsonl b/src/test/resources/xmlWalkerTestDataStreaming/bloomTeragrepStep_data.jsonl
index 732f4fec..2290d262 100644
--- a/src/test/resources/xmlWalkerTestDataStreaming/bloomTeragrepStep_data.jsonl
+++ b/src/test/resources/xmlWalkerTestDataStreaming/bloomTeragrepStep_data.jsonl
@@ -1,2 +1,3 @@
 {"_time": "2022-09-06 09:00:00.000", "id": 1, "_raw": "one", "index": "index_A", "sourcetype": "stream1", "host": "host", "source": "127.0.0.0", "partition": "1", "offset": 1}
-{"_time": "2022-09-06 09:00:00.000", "id": 2, "_raw": "one.two", "index": "index_A", "sourcetype": "stream1", "host": "host", "source": "127.0.0.0", "partition": "2", "offset": 1}
\ No newline at end of file
+{"_time": "2022-09-06 09:00:00.000", "id": 2, "_raw": "one.two", "index": "index_A", "sourcetype": "stream1", "host": "host", "source": "127.0.0.0", "partition": "2", "offset": 1}
+{"_time": "2022-09-06 09:00:00.000", "id": 3, "_raw": "", "index": "index_Empty", "sourcetype": "stream1", "host": "host", "source": "127.0.0.0", "partition": "3", "offset": 1}
\ No newline at end of file