diff --git a/src/fastp.rs b/src/fastp.rs
index d3dbcce..c4f8072 100644
--- a/src/fastp.rs
+++ b/src/fastp.rs
@@ -178,17 +178,29 @@ fn merge_adapter_counts(dicts: &[&Map<String, Value>]) -> Result<Map<String, Val
 
 /// Merge filtering result objects by summing all count fields.
 ///
+/// `adapter_dimer_reads` (added in newer fastp versions) is summed only when present.
+/// If some chunks report it and others don't, `required_i64` errors — mixed-version
+/// inputs fail loudly rather than silently dropping dimer counts.
+///
 /// Errors if any required count field is missing or not an integer.
 fn merge_filtering_result(results: &[Value]) -> Result<Value> {
     const FIELDS: &[&str] = &[
         "passed_filter_reads",
         "low_quality_reads",
         "too_many_N_reads",
+        "adapter_dimer_reads",
         "too_short_reads",
         "too_long_reads",
     ];
+    let has_dimer = results
+        .iter()
+        .any(|r| r.get("adapter_dimer_reads").is_some());
+
     let mut merged = Map::new();
     for field in FIELDS {
+        if *field == "adapter_dimer_reads" && !has_dimer {
+            continue;
+        }
         let mut sum = 0_i64;
         for (i, result) in results.iter().enumerate() {
             sum += required_i64(result, field).with_context(|| format!("filtering_result[{i}]"))?;
diff --git a/tests/fastp_test.rs b/tests/fastp_test.rs
index 045b67f..41e9e54 100644
--- a/tests/fastp_test.rs
+++ b/tests/fastp_test.rs
@@ -191,8 +191,48 @@ fn test_merge_two_fixture_files_filtering_result() {
     assert_eq!(fr["passed_filter_reads"].as_i64(), Some(2850));
     assert_eq!(fr["low_quality_reads"].as_i64(), Some(90));
     assert_eq!(fr["too_many_N_reads"].as_i64(), Some(30));
+    assert_eq!(fr["adapter_dimer_reads"].as_i64(), Some(408));
     assert_eq!(fr["too_short_reads"].as_i64(), Some(30));
     assert_eq!(fr["too_long_reads"].as_i64(), Some(0));
+
+    // adapter_dimer_reads must keep fastp's field order: between too_many_N_reads
+    // and too_short_reads.
+    let keys: Vec<&str> = fr.as_object().unwrap().keys().map(String::as_str).collect();
+    assert_eq!(
+        keys,
+        [
+            "passed_filter_reads",
+            "low_quality_reads",
+            "too_many_N_reads",
+            "adapter_dimer_reads",
+            "too_short_reads",
+            "too_long_reads",
+        ]
+    );
+}
+
+#[test]
+fn test_merge_filtering_result_omits_adapter_dimer_when_absent() {
+    // Older fastp versions predate adapter_dimer_reads; merging must not invent it.
+    let chunk = json!({
+        "summary": {
+            "before_filtering": {"total_reads": 10, "total_bases": 100, "q20_bases": 90, "q30_bases": 80},
+            "after_filtering": {"total_reads": 10, "total_bases": 100, "q20_bases": 90, "q30_bases": 80}
+        },
+        "filtering_result": {
+            "passed_filter_reads": 10,
+            "low_quality_reads": 0,
+            "too_many_N_reads": 0,
+            "too_short_reads": 0,
+            "too_long_reads": 0
+        },
+        "duplication": {"rate": 0.0},
+        "adapter_cutting": {"adapter_trimmed_reads": 0, "adapter_trimmed_bases": 0}
+    });
+    let merged = merge_jsons(&[chunk.clone(), chunk]).expect("merge should succeed");
+    let fr = &merged["filtering_result"];
+    assert!(fr.get("adapter_dimer_reads").is_none());
+    assert_eq!(fr["passed_filter_reads"].as_i64(), Some(20));
 }
 
 #[test]
diff --git a/tests/fixtures/fastp_chunk1.json b/tests/fixtures/fastp_chunk1.json
index af3b700..5872eb8 100644
--- a/tests/fixtures/fastp_chunk1.json
+++ b/tests/fixtures/fastp_chunk1.json
@@ -29,6 +29,7 @@
         "passed_filter_reads": 950,
         "low_quality_reads": 30,
         "too_many_N_reads": 10,
+        "adapter_dimer_reads": 203,
         "too_short_reads": 10,
         "too_long_reads": 0
     },
diff --git a/tests/fixtures/fastp_chunk2.json b/tests/fixtures/fastp_chunk2.json
index 281a13b..5640e74 100644
--- a/tests/fixtures/fastp_chunk2.json
+++ b/tests/fixtures/fastp_chunk2.json
@@ -29,6 +29,7 @@
         "passed_filter_reads": 1900,
         "low_quality_reads": 60,
         "too_many_N_reads": 20,
+        "adapter_dimer_reads": 205,
         "too_short_reads": 20,
         "too_long_reads": 0
     },