diff --git a/HISTORY.md b/HISTORY.md
index 81689fcada..e527d1aa1e 100644
--- a/HISTORY.md
+++ b/HISTORY.md
@@ -4,8 +4,10 @@
 
 _????-??-??_
 
- * Fix compilation with clang 19 (#3799)
+ * Fix compilation with clang 19 (#3799).
 
+ * Deprecate version of `data::Split()` that returns a `std::tuple` for
+   consistency; use other overloads instead (#3803).
 
 ## mlpack 4.5.0
 
diff --git a/src/mlpack/core/data/split_data.hpp b/src/mlpack/core/data/split_data.hpp
index f11956d313..412e871d26 100644
--- a/src/mlpack/core/data/split_data.hpp
+++ b/src/mlpack/core/data/split_data.hpp
@@ -349,6 +349,7 @@ void Split(const MatType& input,
  */
 template<typename T, typename LabelsType,
          typename = std::enable_if_t<arma::is_arma_type<LabelsType>::value> >
+[[deprecated("Will be removed in mlpack 5.0.0; use other overloads instead")]]
 std::tuple<arma::Mat<T>, arma::Mat<T>, LabelsType, LabelsType>
 Split(const arma::Mat<T>& input,
       const LabelsType& inputLabel,
@@ -397,6 +398,7 @@ Split(const arma::Mat<T>& input,
  *      and testData (arma::Mat<T>).
  */
 template<typename T>
+[[deprecated("Will be removed in mlpack 5.0.0; use other overloads instead")]]
 std::tuple<arma::Mat<T>, arma::Mat<T>>
 Split(const arma::Mat<T>& input,
       const double testRatio,
@@ -553,6 +555,7 @@ template <class FieldType, typename T,
           class = std::enable_if_t<
               arma::is_Col<typename FieldType::object_type>::value ||
               arma::is_Mat_only<typename FieldType::object_type>::value>>
+[[deprecated("Will be removed in mlpack 5.0.0; use other overloads instead")]]
 std::tuple<FieldType, FieldType, arma::field<T>, arma::field<T>>
 Split(const FieldType& input,
       const arma::field<T>& inputLabel,
@@ -600,6 +603,7 @@ template <class FieldType,
           class = std::enable_if_t<
               arma::is_Col<typename FieldType::object_type>::value ||
               arma::is_Mat_only<typename FieldType::object_type>::value>>
+[[deprecated("Will be removed in mlpack 5.0.0; use other overloads instead")]]
 std::tuple<FieldType, FieldType>
 Split(const FieldType& input,
       const double testRatio,
diff --git a/src/mlpack/methods/preprocess/preprocess_split_main.cpp b/src/mlpack/methods/preprocess/preprocess_split_main.cpp
index 4410cedb15..5f122a9572 100644
--- a/src/mlpack/methods/preprocess/preprocess_split_main.cpp
+++ b/src/mlpack/methods/preprocess/preprocess_split_main.cpp
@@ -156,40 +156,47 @@ void BINDING_FUNCTION(util::Params& params, util::Timers& timers)
     arma::Row<size_t> labelsRow = labels.row(0);
 
     timers.Start("splitting_data");
-    const auto value =
-        data::Split(data, labelsRow, testRatio, !shuffleData, stratifyData);
+    arma::mat trainData, testData;
+    arma::Row<size_t> trainLabels, testLabels;
+    if (stratifyData)
+    {
+      data::StratifiedSplit(data, labelsRow, trainData, testData, trainLabels,
+          testLabels, testRatio, !shuffleData);
+    }
+    else
+    {
+      data::Split(data, labelsRow, trainData, testData, trainLabels, testLabels,
+          testRatio, !shuffleData);
+    }
     timers.Stop("splitting_data");
 
-    Log::Info << "Training data contains "
-        << get<0>(value).n_cols << " points." << endl;
-    Log::Info << "Test data contains "
-        << get<1>(value).n_cols << " points." << endl;
+    Log::Info << "Training data contains " << trainData.n_cols << " points."
+        << endl;
+    Log::Info << "Test data contains " << testData.n_cols << " points." << endl;
 
     if (params.Has("training"))
-      params.Get<arma::mat>("training") = std::move(get<0>(value));
+      params.Get<arma::mat>("training") = std::move(trainData);
     if (params.Has("test"))
-      params.Get<arma::mat>("test") = std::move(get<1>(value));
+      params.Get<arma::mat>("test") = std::move(testData);
     if (params.Has("training_labels"))
-      params.Get<arma::Mat<size_t>>("training_labels") =
-          std::move(get<2>(value));
+      params.Get<arma::Mat<size_t>>("training_labels") = std::move(trainLabels);
     if (params.Has("test_labels"))
-      params.Get<arma::Mat<size_t>>("test_labels") =
-          std::move(get<3>(value));
+      params.Get<arma::Mat<size_t>>("test_labels") = std::move(testLabels);
   }
   else // We have no labels, so just split the dataset.
   {
     timers.Start("splitting_data");
-    const auto value = data::Split(data, testRatio, !shuffleData);
+    arma::mat trainData, testData;
+    data::Split(data, trainData, testData, testRatio, !shuffleData);
     timers.Stop("splitting_data");
 
-    Log::Info << "Training data contains " << get<0>(value).n_cols << " points."
-        << endl;
-    Log::Info << "Test data contains " << get<1>(value).n_cols << " points."
+    Log::Info << "Training data contains " << trainData.n_cols << " points."
         << endl;
+    Log::Info << "Test data contains " << testData.n_cols << " points." << endl;
 
     if (params.Has("training"))
-      params.Get<arma::mat>("training") = std::move(get<0>(value));
+      params.Get<arma::mat>("training") = std::move(trainData);
     if (params.Has("test"))
-      params.Get<arma::mat>("test") = std::move(get<1>(value));
+      params.Get<arma::mat>("test") = std::move(testData);
   }
 }