diff --git a/doc/source/whatsnew/v1.0.0.rst b/doc/source/whatsnew/v1.0.0.rst
index 7a10447e3ad40..76aadd2c97b67 100644
--- a/doc/source/whatsnew/v1.0.0.rst
+++ b/doc/source/whatsnew/v1.0.0.rst
@@ -187,7 +187,7 @@ Reshaping
 
 Sparse
 ^^^^^^
-
+- Bug in :class:`SparseDataFrame` arithmetic operations incorrectly casting inputs to float (:issue:`28107`)
 -
 -
 
diff --git a/pandas/core/sparse/frame.py b/pandas/core/sparse/frame.py
index 8fe6850c84b8b..3d6ba0b8d9774 100644
--- a/pandas/core/sparse/frame.py
+++ b/pandas/core/sparse/frame.py
@@ -576,8 +576,8 @@ def _combine_match_index(self, other, func, level=None):
         this, other = self.align(other, join="outer", axis=0, level=level, copy=False)
 
         new_data = {}
-        for col, series in this.items():
-            new_data[col] = func(series.values, other.values)
+        for col in this.columns:
+            new_data[col] = func(this[col], other)
 
         fill_value = self._get_op_result_fill_value(other, func)
 
@@ -603,7 +603,7 @@ def _combine_match_columns(self, other, func, level=None):
         new_data = {}
 
         for col in left.columns:
-            new_data[col] = func(left[col], float(right[col]))
+            new_data[col] = func(left[col], right[col])
 
         return self._constructor(
             new_data,
diff --git a/pandas/tests/sparse/frame/test_frame.py b/pandas/tests/sparse/frame/test_frame.py
index ddb50e0897a86..e372e2563e682 100644
--- a/pandas/tests/sparse/frame/test_frame.py
+++ b/pandas/tests/sparse/frame/test_frame.py
@@ -1487,6 +1487,22 @@ def test_comparison_op_scalar(self):
         assert isinstance(res, pd.SparseDataFrame)
         tm.assert_frame_equal(res.to_dense(), df != 0)
 
+    def test_add_series_retains_dtype(self):
+        # SparseDataFrame._combine_match_columns used to incorrectly cast
+        #  to float
+        d = {0: [2j, 3j], 1: [0, 1]}
+        sdf = SparseDataFrame(data=d, default_fill_value=1)
+        result = sdf + sdf[0]
+
+        df = sdf.to_dense()
+        expected = df + df[0]
+        tm.assert_frame_equal(result.to_dense(), expected)
+
+        # Make it explicit to be on the safe side
+        edata = {0: [4j, 5j], 1: [3j, 1 + 3j]}
+        expected = DataFrame(edata)
+        tm.assert_frame_equal(result.to_dense(), expected)
+
 
 @pytest.mark.filterwarnings("ignore:Sparse:FutureWarning")
 @pytest.mark.filterwarnings("ignore:DataFrame.to_sparse:FutureWarning")