diff --git a/dbms/include/DB/Storages/MergeTree/MergedBlockOutputStream.h b/dbms/include/DB/Storages/MergeTree/MergedBlockOutputStream.h
index fb66743d71e..c50f6b3a79c 100644
--- a/dbms/include/DB/Storages/MergeTree/MergedBlockOutputStream.h
+++ b/dbms/include/DB/Storages/MergeTree/MergedBlockOutputStream.h
@@ -78,6 +78,11 @@ protected:
 	size_t aio_threshold;
 
 	CompressionMethod compression_method;
+
+private:
+	/// Internal version of writeData.
+	void writeDataImpl(const String & name, const IDataType & type, const IColumn & column,
+		OffsetColumns & offset_columns, size_t level, bool write_array_data);
 };
 
 
diff --git a/dbms/src/Storages/MergeTree/MergedBlockOutputStream.cpp b/dbms/src/Storages/MergeTree/MergedBlockOutputStream.cpp
index a2c834faa87..2d72960ffac 100644
--- a/dbms/src/Storages/MergeTree/MergedBlockOutputStream.cpp
+++ b/dbms/src/Storages/MergeTree/MergedBlockOutputStream.cpp
@@ -99,7 +99,19 @@ void IMergedBlockOutputStream::addStream(const String & path, const String & nam
 /// Записать данные одного столбца.
 void IMergedBlockOutputStream::writeData(const String & name, const IDataType & type, const IColumn & column, OffsetColumns & offset_columns, size_t level)
 {
+	writeDataImpl(name, type, column, offset_columns, level, false);
+}
+
+void IMergedBlockOutputStream::writeDataImpl(const String & name, const IDataType & type,
+	const IColumn & column, OffsetColumns & offset_columns, size_t level, bool write_array_data)
+{
+	/// NOTE: the parameter write_array_data indicates whether we call this method
+	/// to write the contents of an array. This is to cope with the fact that
+	/// serialization of arrays for the MergeTree engine slightly differs from
+	/// what the other engines do.
+
 	size_t size = column.size();
+	const DataTypeArray * type_arr;
 
 	if (type.isNullable())
 	{
@@ -120,9 +132,7 @@ void IMergedBlockOutputStream::writeData(const String & name, const IDataType &
 
 			/// Если есть index_offset, то первая засечка идёт не сразу, а после этого количества строк.
 			if (prev_mark == 0 && index_offset != 0)
-			{
 				limit = index_offset;
-			}
 			else
 			{
 				limit = storage.index_granularity;
@@ -144,9 +154,9 @@ void IMergedBlockOutputStream::writeData(const String & name, const IDataType &
 		}
 
 		/// Then write data.
-		writeData(name, nested_type, nested_col, offset_columns, level);
+		writeDataImpl(name, nested_type, nested_col, offset_columns, level, write_array_data);
 	}
-	else if (const DataTypeArray * type_arr = typeid_cast<const DataTypeArray *>(&type))
+	else if (!write_array_data && ((type_arr = typeid_cast<const DataTypeArray *>(&type)) != nullptr))
 	{
 		/// Для массивов требуется сначала сериализовать размеры, а потом значения.
 		String size_name = DataTypeNested::extractNestedTableName(name)
@@ -165,9 +175,7 @@ void IMergedBlockOutputStream::writeData(const String & name, const IDataType &
 
 				/// Если есть index_offset, то первая засечка идёт не сразу, а после этого количества строк.
 				if (prev_mark == 0 && index_offset != 0)
-				{
 					limit = index_offset;
-				}
 				else
 				{
 					limit = storage.index_granularity;
@@ -189,7 +197,12 @@ void IMergedBlockOutputStream::writeData(const String & name, const IDataType &
 			}
 		}
 
-		writeData(name, *type_arr->getNestedType(), typeid_cast<const ColumnArray &>(column).getData(), offset_columns, level + 1);
+		if (type_arr->getNestedType()->isNullable())
+			writeDataImpl(name, *type_arr->getNestedType(),
+				typeid_cast<const ColumnArray &>(column).getData(), offset_columns,
+				level + 1, true);
+		else
+			writeDataImpl(name, type, column, offset_columns, level + 1, true);
 	}
 	else
 	{
@@ -202,9 +215,7 @@ void IMergedBlockOutputStream::writeData(const String & name, const IDataType &
 
 			/// Если есть index_offset, то первая засечка идёт не сразу, а после этого количества строк.
 			if (prev_mark == 0 && index_offset != 0)
-			{
 				limit = index_offset;
-			}
 			else
 			{
 				limit = storage.index_granularity;