diff --git a/src/query/src/part_sort.rs b/src/query/src/part_sort.rs
index d44304b415..f1574ec334 100644
--- a/src/query/src/part_sort.rs
+++ b/src/query/src/part_sort.rs
@@ -534,43 +534,69 @@ impl PartSortStream {
         }
 
         if topk {
-            self.compact_topk_buffer()?;
-            self.update_dynamic_filter(sort_data_type)?;
+            let threshold = self.compact_topk_buffer(sort_data_type)?;
+            self.update_dynamic_filter(sort_data_type, threshold)?;
         }
 
         Ok(())
     }
 
-    fn compact_topk_buffer(&mut self) -> datafusion_common::Result<()> {
+    fn compact_topk_buffer(
+        &mut self,
+        sort_data_type: &DataType,
+    ) -> datafusion_common::Result<Option<TopKThreshold>> {
         let Some(limit) = self.limit else {
-            return Ok(());
+            return Ok(None);
         };
 
         let PartSortBuffer::TopK(buffer) =
             std::mem::replace(&mut self.buffer, PartSortBuffer::TopK(Vec::new()))
         else {
-            return Ok(());
+            return Ok(None);
         };
 
         if limit == 0 || buffer.is_empty() {
             self.buffer = PartSortBuffer::TopK(Vec::new());
-            return Ok(());
+            return Ok(None);
         }
 
         let total_rows: usize = buffer.iter().map(|batch| batch.num_rows()).sum();
         if total_rows <= limit {
             self.buffer = PartSortBuffer::TopK(buffer);
-            return Ok(());
+            return Ok(None);
         }
 
         let topk = self.sort_record_batches(&buffer, Some(limit), false)?;
+        let threshold = self.threshold_from_sorted_batch(&topk, sort_data_type)?;
         self.buffer = if topk.num_rows() == 0 {
             PartSortBuffer::TopK(Vec::new())
         } else {
             PartSortBuffer::TopK(vec![topk])
         };
 
-        Ok(())
+        Ok(threshold)
+    }
+
+    fn threshold_from_sorted_batch(
+        &self,
+        batch: &DfRecordBatch,
+        sort_data_type: &DataType,
+    ) -> datafusion_common::Result<Option<TopKThreshold>> {
+        if batch.num_rows() == 0 {
+            return Ok(None);
+        }
+
+        let threshold_idx = batch.num_rows() - 1;
+        let sort_column = self.expression.evaluate_to_sort_column(batch)?.values;
+        let threshold = downcast_ts_array!(
+            sort_data_type => (threshold_helper, sort_column, threshold_idx),
+            _ => internal_err!(
+                "Unsupported data type for sort column: {:?}",
+                sort_data_type
+            )?,
+        );
+
+        Ok(Some(threshold))
     }
 
     fn topk_threshold(
@@ -581,7 +607,7 @@ impl PartSortStream {
             return Ok(None);
         };
 
-        if self.buffer.num_rows() < limit {
+        if limit == 0 || self.buffer.num_rows() < limit {
             return Ok(None);
         }
 
@@ -691,13 +717,19 @@ impl PartSortStream {
     fn update_dynamic_filter(
         &mut self,
         sort_data_type: &DataType,
+        threshold: Option<TopKThreshold>,
     ) -> datafusion_common::Result<()> {
         let Some(filter) = &self.dynamic_filter else {
             return Ok(());
         };
 
-        let Some(threshold) = self.topk_threshold(sort_data_type)? else {
-            return Ok(());
+        let threshold = if let Some(threshold) = threshold {
+            threshold
+        } else {
+            let Some(threshold) = self.topk_threshold(sort_data_type)? else {
+                return Ok(());
+            };
+            threshold
         };
 
         if self.dynamic_filter_threshold.as_ref() == Some(&threshold) {
@@ -722,8 +754,13 @@ impl PartSortStream {
             return Ok(false);
         }
 
-        let Some(threshold) = self.topk_threshold(sort_data_type)? else {
-            return Ok(false);
+        let threshold = if let Some(threshold) = &self.dynamic_filter_threshold {
+            threshold.clone()
+        } else {
+            let Some(threshold) = self.topk_threshold(sort_data_type)? else {
+                return Ok(false);
+            };
+            threshold
         };
 
         let (_, start_idx, _) = self.range_groups[group_idx];
@@ -1881,6 +1918,59 @@ mod test {
         assert_eq!(stream.sort_buffer().unwrap(), expected);
     }
 
+    #[test]
+    fn test_topk_limit_zero_clears_buffer_without_threshold() {
+        let unit = TimeUnit::Millisecond;
+        let schema = Arc::new(Schema::new(vec![Field::new(
+            "ts",
+            DataType::Timestamp(unit, None),
+            false,
+        )]));
+        let sort_data_type = DataType::Timestamp(unit, None);
+        let partition_range = PartitionRange {
+            start: Timestamp::new(0, unit.into()),
+            end: Timestamp::new(10, unit.into()),
+            num_rows: 3,
+            identifier: 0,
+        };
+        let mock_input = Arc::new(MockInputExec::new(vec![vec![]], schema.clone()));
+        let exec = PartSortExec::try_new(
+            PhysicalSortExpr {
+                expr: Arc::new(Column::new("ts", 0)),
+                options: SortOptions {
+                    descending: true,
+                    ..Default::default()
+                },
+            },
+            Some(0),
+            vec![vec![partition_range]],
+            mock_input.clone(),
+        )
+        .unwrap();
+        let input_stream = mock_input
+            .execute(0, Arc::new(TaskContext::default()))
+            .unwrap();
+        let mut stream = PartSortStream::new(
+            Arc::new(TaskContext::default()),
+            &exec,
+            Some(0),
+            input_stream,
+            vec![partition_range],
+            0,
+        )
+        .unwrap();
+
+        stream
+            .push_buffer(
+                DfRecordBatch::try_new(schema, vec![new_ts_array(unit, vec![1, 2, 3])]).unwrap(),
+                &sort_data_type,
+            )
+            .unwrap();
+
+        assert_eq!(stream.buffer.num_rows(), 0);
+        assert_eq!(stream.dynamic_filter_threshold, None);
+    }
+
     /// Test that verifies early termination behavior.
     /// Once we've produced limit * num_partitions rows, we should stop
     /// pulling from input stream.