diff --git a/python/paddle/fluid/tests/unittests/test_dist_transpiler.py b/python/paddle/fluid/tests/unittests/test_dist_transpiler.py
index 0e44cee48b9990163b3bee1488e8f908a91678c9..986fdd9ff27fe2be54ce97f330028b4ae2358714 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_transpiler.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_transpiler.py
@@ -283,6 +283,25 @@ class TestDecayedAdagrad(TranspilerTest):
         trainer, _ = self.get_trainer()
 
 
+class TestFtrl(TranspilerTest):
+    def net_conf(self):
+        x = fluid.layers.data(name='x', shape=[1000], dtype='float32')
+        y_predict = fluid.layers.fc(input=x,
+                                    size=1000,
+                                    act=None,
+                                    param_attr=fluid.ParamAttr(name='fc_w'),
+                                    bias_attr=fluid.ParamAttr(name='fc_b'))
+        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        cost = fluid.layers.square_error_cost(input=y_predict, label=y)
+        avg_cost = fluid.layers.mean(cost)
+        opt = fluid.optimizer.Ftrl(learning_rate=0.1)
+        opt.minimize(avg_cost)
+
+    def transpiler_test_impl(self):
+        pserver, startup = self.get_pserver(self.pserver1_ep)
+        trainer, _ = self.get_trainer()
+
+
 class TestLRDecayConditional(TranspilerTest):
     def net_conf(self):
         x = fluid.layers.data(name='x', shape=[1000], dtype='float32')
diff --git a/python/paddle/fluid/transpiler/distribute_transpiler.py b/python/paddle/fluid/transpiler/distribute_transpiler.py
index 17a8720f5266c14829702c08aac8eeffbd988b07..4af13b605fa7054df097e3bf0ed0c71f468f02de 100644
--- a/python/paddle/fluid/transpiler/distribute_transpiler.py
+++ b/python/paddle/fluid/transpiler/distribute_transpiler.py
@@ -1456,6 +1456,9 @@ to transpile() call.")
         elif op_type == "decayed_adagrad":
             if varkey == "Moment":
                 return param_shape
+        elif op_type == "ftrl":
+            if varkey in ["SquaredAccumulator", "LinearAccumulator"]:
+                return param_shape
         elif op_type == "sgd":
             pass
         else: