diff --git a/Makefile b/Makefile
index d853692..160eb95 100644
--- a/Makefile
+++ b/Makefile
@@ -78,10 +78,13 @@ data_pretrain: requirements
 data_pretrain_cv: requirements
 	$(PYTHON_INTERPRETER) scripts/process_external_cv.py
 
-## Process internal data with amine-based CV splitting (interim -> processed/cv)
+## Process internal data with CV splitting (interim -> processed/cv)
+## Use SCAFFOLD_SPLIT=1 to enable amine-based scaffold splitting (default: random shuffle)
+SCAFFOLD_SPLIT_FLAG = $(if $(filter 1,$(SCAFFOLD_SPLIT)),--scaffold-split,)
+
 .PHONY: data_cv
 data_cv: requirements
-	$(PYTHON_INTERPRETER) scripts/process_data_cv.py
+	$(PYTHON_INTERPRETER) scripts/process_data_cv.py $(SCAFFOLD_SPLIT_FLAG)
 
 # MPNN 支持：使用 USE_MPNN=1 启用 MPNN encoder
 # 例如：make pretrain USE_MPNN=1
diff --git a/data/processed/cv/fold_0/test.parquet b/data/processed/cv/fold_0/test.parquet
index 7799d90..a5f5215 100644
Binary files a/data/processed/cv/fold_0/test.parquet and b/data/processed/cv/fold_0/test.parquet differ
diff --git a/data/processed/cv/fold_0/train.parquet b/data/processed/cv/fold_0/train.parquet
index e753b68..e3c307e 100644
Binary files a/data/processed/cv/fold_0/train.parquet and b/data/processed/cv/fold_0/train.parquet differ
diff --git a/data/processed/cv/fold_0/val.parquet b/data/processed/cv/fold_0/val.parquet
index 1f14c35..7905976 100644
Binary files a/data/processed/cv/fold_0/val.parquet and b/data/processed/cv/fold_0/val.parquet differ
diff --git a/data/processed/cv/fold_1/test.parquet b/data/processed/cv/fold_1/test.parquet
index 30b25a8..14f37b7 100644
Binary files a/data/processed/cv/fold_1/test.parquet and b/data/processed/cv/fold_1/test.parquet differ
diff --git a/data/processed/cv/fold_1/train.parquet b/data/processed/cv/fold_1/train.parquet
index c096898..1cfa625 100644
Binary files a/data/processed/cv/fold_1/train.parquet and b/data/processed/cv/fold_1/train.parquet differ
diff --git a/data/processed/cv/fold_1/val.parquet b/data/processed/cv/fold_1/val.parquet
index 7799d90..a5f5215 100644
Binary files a/data/processed/cv/fold_1/val.parquet and b/data/processed/cv/fold_1/val.parquet differ
diff --git a/data/processed/cv/fold_2/test.parquet b/data/processed/cv/fold_2/test.parquet
index b3bb10b..5605067 100644
Binary files a/data/processed/cv/fold_2/test.parquet and b/data/processed/cv/fold_2/test.parquet differ
diff --git a/data/processed/cv/fold_2/train.parquet b/data/processed/cv/fold_2/train.parquet
index bef51fc..31e66b6 100644
Binary files a/data/processed/cv/fold_2/train.parquet and b/data/processed/cv/fold_2/train.parquet differ
diff --git a/data/processed/cv/fold_2/val.parquet b/data/processed/cv/fold_2/val.parquet
index 30b25a8..14f37b7 100644
Binary files a/data/processed/cv/fold_2/val.parquet and b/data/processed/cv/fold_2/val.parquet differ
diff --git a/data/processed/cv/fold_3/test.parquet b/data/processed/cv/fold_3/test.parquet
index 93357c0..ed3b26d 100644
Binary files a/data/processed/cv/fold_3/test.parquet and b/data/processed/cv/fold_3/test.parquet differ
diff --git a/data/processed/cv/fold_3/train.parquet b/data/processed/cv/fold_3/train.parquet
index b215888..ec51f26 100644
Binary files a/data/processed/cv/fold_3/train.parquet and b/data/processed/cv/fold_3/train.parquet differ
diff --git a/data/processed/cv/fold_3/val.parquet b/data/processed/cv/fold_3/val.parquet
index b3bb10b..5605067 100644
Binary files a/data/processed/cv/fold_3/val.parquet and b/data/processed/cv/fold_3/val.parquet differ
diff --git a/data/processed/cv/fold_4/test.parquet b/data/processed/cv/fold_4/test.parquet
index 1f14c35..7905976 100644
Binary files a/data/processed/cv/fold_4/test.parquet and b/data/processed/cv/fold_4/test.parquet differ
diff --git a/data/processed/cv/fold_4/train.parquet b/data/processed/cv/fold_4/train.parquet
index 179480e..c802f21 100644
Binary files a/data/processed/cv/fold_4/train.parquet and b/data/processed/cv/fold_4/train.parquet differ
diff --git a/data/processed/cv/fold_4/val.parquet b/data/processed/cv/fold_4/val.parquet
index 93357c0..ed3b26d 100644
Binary files a/data/processed/cv/fold_4/val.parquet and b/data/processed/cv/fold_4/val.parquet differ
diff --git a/scripts/process_data_cv.py b/scripts/process_data_cv.py
index 7be689e..1b02404 100644
--- a/scripts/process_data_cv.py
+++ b/scripts/process_data_cv.py
@@ -1,4 +1,4 @@
-"""内部数据 Cross-Validation 划分脚本：基于 Amine 的分组划分"""
+"""内部数据 Cross-Validation 划分脚本：支持随机划分或基于 Amine 的分组划分"""
 
 from pathlib import Path
 from typing import List
@@ -26,6 +26,71 @@ from lnp_ml.dataset import (
 app = typer.Typer()
 
 
+def random_cv_split(
+    df: pd.DataFrame,
+    n_folds: int = 5,
+    seed: int = 42,
+) -> List[dict]:
+    """
+    随机 shuffle 进行 Cross-Validation 划分。
+    
+    步骤：
+    1. 打乱所有样本
+    2. 将样本分成 n_folds 个容器
+    3. 对于每个 fold i：
+       - validation = container[i]
+       - test = container[(i+1) % n_folds]
+       - train = 其余所有
+    
+    Args:
+        df: 输入 DataFrame
+        n_folds: 折数
+        seed: 随机种子
+        
+    Returns:
+        List of dicts，每个 dict 包含 train_df, val_df, test_df
+    """
+    # 打乱所有样本
+    df_shuffled = df.sample(frac=1, random_state=seed).reset_index(drop=True)
+    n_samples = len(df_shuffled)
+    
+    logger.info(f"Total {n_samples} samples for random CV split")
+    
+    # 将样本分成 n_folds 个容器
+    indices = np.arange(n_samples)
+    containers = np.array_split(indices, n_folds)
+    
+    # 打印每个容器的大小
+    for i, container in enumerate(containers):
+        logger.info(f"  Container {i}: {len(container)} samples")
+    
+    # 生成每个 fold 的数据
+    fold_splits = []
+    for i in range(n_folds):
+        val_indices = containers[i]
+        test_indices = containers[(i + 1) % n_folds]
+        train_indices = np.concatenate([
+            containers[j] for j in range(n_folds) 
+            if j != i and j != (i + 1) % n_folds
+        ])
+        
+        train_df = df_shuffled.iloc[train_indices].reset_index(drop=True)
+        val_df = df_shuffled.iloc[val_indices].reset_index(drop=True)
+        test_df = df_shuffled.iloc[test_indices].reset_index(drop=True)
+        
+        fold_splits.append({
+            "train": train_df,
+            "val": val_df,
+            "test": test_df,
+        })
+        
+        logger.info(
+            f"Fold {i}: train={len(train_df)}, val={len(val_df)}, test={len(test_df)}"
+        )
+    
+    return fold_splits
+
+
 def amine_based_cv_split(
     df: pd.DataFrame,
     n_folds: int = 5,
@@ -106,12 +171,18 @@ def main(
     n_folds: int = 5,
     seed: int = 42,
     amine_col: str = "Amine",
+    scaffold_split: bool = typer.Option(
+        False,
+        "--scaffold-split",
+        help="使用基于 Amine 的 scaffold splitting（默认：随机 shuffle）",
+    ),
 ):
     """
-    基于 Amine 分组进行 Cross-Validation 数据划分。
+    Cross-Validation 数据划分。
     
-    采用类似 scaffold splitting 的思路，将相同 Amine 的数据放在同一组，
-    确保训练集和测试集之间没有 Amine 泄露。
+    支持两种划分方式：
+    - 随机划分（默认）：直接 shuffle 所有样本
+    - Scaffold splitting（--scaffold-split）：基于 Amine 分组，确保同一 Amine 的数据在同一组
     
     划分比例约为 train:val:test ≈ 3:1:1
     
@@ -126,20 +197,19 @@ def main(
     df = pd.read_csv(input_path)
     logger.info(f"Loaded {len(df)} samples")
     
-    # 检查 amine 列是否存在
-    if amine_col not in df.columns:
-        logger.error(f"Column '{amine_col}' not found in data. Available columns: {list(df.columns)}")
-        raise typer.Exit(1)
-    
     # 处理数据（列对齐、one-hot 生成等）
     logger.info("Processing dataframe...")
     df = process_dataframe(df)
     
-    # 确保 Amine 列仍然存在（process_dataframe 可能不会保留它）
-    # 重新加载原始数据获取 Amine 列
-    original_df = pd.read_csv(input_path)
-    if amine_col in original_df.columns and amine_col not in df.columns:
-        df[amine_col] = original_df[amine_col].values
+    # 如果使用 scaffold split，检查 amine 列是否存在
+    if scaffold_split:
+        # 重新加载原始数据获取 Amine 列（process_dataframe 可能不会保留它）
+        original_df = pd.read_csv(input_path)
+        if amine_col not in original_df.columns:
+            logger.error(f"Column '{amine_col}' not found in data. Available columns: {list(original_df.columns)}")
+            raise typer.Exit(1)
+        if amine_col not in df.columns:
+            df[amine_col] = original_df[amine_col].values
     
     # 定义要保留的列
     phys_cols = get_phys_cols()
@@ -162,8 +232,14 @@ def main(
     keep_cols = [c for c in keep_cols if c in df.columns]
     
     # 进行 CV 划分
-    logger.info(f"\nPerforming {n_folds}-fold amine-based CV split (seed={seed})...")
-    fold_splits = amine_based_cv_split(df, n_folds=n_folds, seed=seed, amine_col=amine_col)
+    if scaffold_split:
+        logger.info(f"\nPerforming {n_folds}-fold amine-based scaffold CV split (seed={seed})...")
+        fold_splits = amine_based_cv_split(df, n_folds=n_folds, seed=seed, amine_col=amine_col)
+        split_method = f"Amine-based scaffold (column: {amine_col})"
+    else:
+        logger.info(f"\nPerforming {n_folds}-fold random CV split (seed={seed})...")
+        fold_splits = random_cv_split(df, n_folds=n_folds, seed=seed)
+        split_method = "Random shuffle"
     
     # 保存每个 fold
     output_dir.mkdir(parents=True, exist_ok=True)
@@ -217,7 +293,7 @@ def main(
     logger.info("=" * 60)
     logger.info(f"Output directory: {output_dir}")
     logger.info(f"Number of folds: {n_folds}")
-    logger.info(f"Splitting method: Amine-based (column: {amine_col})")
+    logger.info(f"Splitting method: {split_method}")
     logger.info(f"Random seed: {seed}")