diff --git a/dataduct/etl/etl_pipeline.py b/dataduct/etl/etl_pipeline.py
index 5cd5aed..19bc1d1 100644
--- a/dataduct/etl/etl_pipeline.py
+++ b/dataduct/etl/etl_pipeline.py
@@ -60,6 +60,11 @@ class ETLPipeline(object):
     and has functionality to add steps to the pipeline
 
     """
+
+    # put this here so as not to pollute global namespace. Also makes mocking
+    # easier
+    DEFAULT_TOPIC_ARN = config.etl.get('DEFAULT_TOPIC_ARN', const.NONE)
+
     def __init__(self, name, frequency='one-time', ec2_resource_config=None,
                  time_delta=None, emr_cluster_config=None, load_time=None,
                  topic_arn=None, max_retries=MAX_RETRIES, teardown=None,
@@ -93,7 +98,13 @@ def __init__(self, name, frequency='one-time', ec2_resource_config=None,
         self.time_delta = time_delta
         self.description = description
         self.max_retries = max_retries
-        self.topic_arn = topic_arn
+
+        if topic_arn is not None:
+            self.topic_arn = topic_arn
+        elif self.DEFAULT_TOPIC_ARN:
+            self.topic_arn = self.DEFAULT_TOPIC_ARN
+        else:
+            self.topic_arn = None
 
         if bootstrap is not None:
             self.bootstrap_definitions = bootstrap
diff --git a/dataduct/etl/tests/test_etl_pipeline.py b/dataduct/etl/tests/test_etl_pipeline.py
index a40c16c..28e135e 100644
--- a/dataduct/etl/tests/test_etl_pipeline.py
+++ b/dataduct/etl/tests/test_etl_pipeline.py
@@ -1,14 +1,15 @@
 """Tests for the ETL Pipeline object
 """
 import unittest
+import mock
 from nose.tools import raises
 from nose.tools import eq_
+from nose.tools import assert_not_equal
 
 from datetime import timedelta
 from ..etl_pipeline import ETLPipeline
 from ...utils.exceptions import ETLInputError
 
-
 class EtlPipelineTests(unittest.TestCase):
     """Tests for the ETL Pipeline object
     """
@@ -59,3 +60,16 @@ def test_bad_data_type_throws(self):
         _s3_uri is bad
         """
         self.default_pipeline._s3_uri('TEST_DATA_TYPE')
+
+    @staticmethod
+    def test_default_arn_loaded_if_not_in_etl_yaml():
+        with mock.patch('dataduct.etl.etl_pipeline.ETLPipeline.DEFAULT_TOPIC_ARN', 'blah'):
+            result = ETLPipeline('test_pipeline')
+            eq_(result.topic_arn, 'blah')
+
+    @staticmethod
+    def test_arn_loads_if_provided_in_etl_yaml():
+        with mock.patch('dataduct.etl.etl_pipeline.ETLPipeline.DEFAULT_TOPIC_ARN', 'blah'):
+            result = ETLPipeline('test_pipeline', topic_arn="not_blah")
+            eq_(result.topic_arn, "not_blah")
+            assert_not_equal(result.topic_arn, ETLPipeline.DEFAULT_TOPIC_ARN)
diff --git a/docs/config.rst b/docs/config.rst
index 3f7401a..3cb957f 100644
--- a/docs/config.rst
+++ b/docs/config.rst
@@ -181,6 +181,7 @@ ETL
         ROLE: FILL_ME_IN
         S3_BASE_PATH: dev
         S3_ETL_BUCKET: FILL_ME_IN
+        DEFAULT_TOPIC_ARN: 'arn:aws:sns:example_arn'
         SNS_TOPIC_ARN_FAILURE: null
         SNS_TOPIC_ARN_WARNING: null
         FREQUENCY_OVERRIDE: one-time
@@ -222,6 +223,8 @@ level. The parameters are explained below:
    or across production and dev
 -  ``S3_ETL_BUCKET``: S3 bucket to use for DP data, logs, source code
    etc.
+-  ``DEFAULT_TOPIC_ARN``: default ARN to use for pipelines. Overridden if
+   specified in pipeline yaml definition.
 -  ``SNS_TOPIC_ARN_FAILURE``: SNS to trigger for failed steps or
    pipelines
 -  ``SNS_TOPIC_ARN_WARNING``: SNS to trigger for failed QA checks