diff --git a/sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveSessionCatalog.scala b/sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveSessionCatalog.scala
index 8a47dcf90803..2589b9d4a028 100644
--- a/sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveSessionCatalog.scala
+++ b/sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveSessionCatalog.scala
@@ -30,12 +30,13 @@ import org.apache.spark.sql.catalyst.{FunctionIdentifier, TableIdentifier}
 import org.apache.spark.sql.catalyst.analysis.FunctionRegistry
 import org.apache.spark.sql.catalyst.analysis.FunctionRegistry.FunctionBuilder
 import org.apache.spark.sql.catalyst.catalog.{FunctionResourceLoader, SessionCatalog}
-import org.apache.spark.sql.catalyst.expressions.{Expression, ExpressionInfo}
+import org.apache.spark.sql.catalyst.expressions.{Cast, Expression, ExpressionInfo}
 import org.apache.spark.sql.catalyst.plans.logical.{LogicalPlan, SubqueryAlias}
 import org.apache.spark.sql.catalyst.rules.Rule
 import org.apache.spark.sql.hive.HiveShim.HiveFunctionWrapper
 import org.apache.spark.sql.hive.client.HiveClient
 import org.apache.spark.sql.internal.SQLConf
+import org.apache.spark.sql.types.{DecimalType, DoubleType}
 import org.apache.spark.util.Utils
 
 
@@ -163,6 +164,19 @@ private[sql] class HiveSessionCatalog(
   }
 
   override def lookupFunction(name: FunctionIdentifier, children: Seq[Expression]): Expression = {
+    try {
+      lookupFunction0(name, children)
+    } catch {
+      case NonFatal(_) =>
+        // SPARK-16228 ExternalCatalog may recognize `double`-type only.
+        val newChildren = children.map { child =>
+          if (child.dataType.isInstanceOf[DecimalType]) Cast(child, DoubleType) else child
+        }
+        lookupFunction0(name, newChildren)
+    }
+  }
+
+  private def lookupFunction0(name: FunctionIdentifier, children: Seq[Expression]): Expression = {
     // TODO: Once lookupFunction accepts a FunctionIdentifier, we should refactor this method to
     // if (super.functionExists(name)) {
     //   super.lookupFunction(name, children)
diff --git a/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveUDFSuite.scala b/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveUDFSuite.scala
index 0f56b2c0d1f4..def4601cf615 100644
--- a/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveUDFSuite.scala
+++ b/sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveUDFSuite.scala
@@ -142,6 +142,13 @@ class HiveUDFSuite extends QueryTest with TestHiveSingleton with SQLTestUtils {
       sql("SELECT array(max(key), max(key)) FROM src").collect().toSeq)
   }
 
+  test("SPARK-16228 Percentile needs explicit cast to double") {
+    sql("select percentile(value, cast(0.5 as double)) from values 1,2,3 T(value)")
+    sql("select percentile_approx(value, cast(0.5 as double)) from values 1.0,2.0,3.0 T(value)")
+    sql("select percentile(value, 0.5) from values 1,2,3 T(value)")
+    sql("select percentile_approx(value, 0.5) from values 1.0,2.0,3.0 T(value)")
+  }
+
   test("Generic UDAF aggregates") {
     checkAnswer(sql("SELECT ceiling(percentile_approx(key, 0.99999D)) FROM src LIMIT 1"),
       sql("SELECT max(key) FROM src LIMIT 1").collect().toSeq)