diff --git a/datapusher/jobs.py b/datapusher/jobs.py
index 18bb01a9..3eda299f 100644
--- a/datapusher/jobs.py
+++ b/datapusher/jobs.py
@@ -66,11 +66,16 @@
 if not SSL_VERIFY:
     requests.packages.urllib3.disable_warnings()
 
+POSTGRES_INT_MAX = 2147483647
+POSTGRES_INT_MIN = -2147483648
+POSTGRES_BIGINT_MAX = 9223372036854775807
+POSTGRES_BIGINT_MIN = -9223372036854775808
+
 _TYPE_MAPPING = {
     'String': 'text',
-    # 'int' may not be big enough,
-    # and type detection may not realize it needs to be big
-    'Integer': 'numeric',
+    # smartint looks at the min/max values of qsv stats scan and
+    # chooses integer, bigint or numeric as appropriate
+    'Integer': 'smartint',
     'Float': 'numeric',
     'DateTime': 'timestamp',
     'Date': 'date',
@@ -586,6 +591,8 @@ def push_to_datastore(task_id, input, dry_run=False):
     # run qsv stats to get data types and descriptive statistics
     headers = []
     types = []
+    headers_min = []
+    headers_max = []
     qsv_stats_csv = tempfile.NamedTemporaryFile(suffix='.csv')
     qsv_stats_cmd = [QSV_BIN, 'stats', tmp.name,
                      '--output', qsv_stats_csv.name]
@@ -605,11 +612,12 @@ def push_to_datastore(task_id, input, dry_run=False):
             'Cannot infer data types and compile statistics: {}'.format(e)
         )
     with open(qsv_stats_csv.name, mode='r') as inp:
-        reader = csv.reader(inp)
-        next(reader)  # skip first element, which is a header
-        for rows in reader:
-            headers.append(rows[0])
-            types.append(rows[1])
+        reader = csv.DictReader(inp)
+        for row in reader:
+            headers.append(row['field'])
+            types.append(row['type'])
+            headers_min.append(row['min'])
+            headers_max.append(row['max'])
 
     existing = datastore_resource_exists(resource_id, api_key, ckan_url)
     existing_info = None
@@ -621,7 +629,7 @@ def push_to_datastore(task_id, input, dry_run=False):
     if existing_info:
         types = [{
             'text': 'String',
-            'numeric': 'Decimal',
+            'numeric': 'Float',
             'timestamp': 'DateTime',
         }.get(existing_info.get(h, {}).get('type_override'), t)
             for t, h in zip(types, headers)]
@@ -636,8 +644,23 @@ def push_to_datastore(task_id, input, dry_run=False):
             res_id=resource_id))
         delete_datastore_resource(resource_id, api_key, ckan_url)
 
-    headers_dicts = [dict(id=field[0], type=TYPE_MAPPING[str(field[1])])
-                     for field in zip(headers, types)]
+    # 1st pass of building headers_dict
+    temp_headers_dicts = [dict(id=field[0], type=TYPE_MAPPING[str(field[1])])
+                          for field in zip(headers, types)]
+
+    # 2nd pass header_dicts, checking for smartint types
+    headers_dicts = []
+    for idx, header in enumerate(temp_headers_dicts):
+        if header['type'] == 'smartint':
+            if int(headers_max[idx]) <= POSTGRES_INT_MAX and int(headers_min[idx]) >= POSTGRES_INT_MIN:
+                header_type = 'integer'
+            elif int(headers_max[idx]) <= POSTGRES_BIGINT_MAX and int(headers_min[idx]) >= POSTGRES_BIGINT_MIN:
+                header_type = 'bigint'
+            else:
+                header_type = 'numeric'
+        else:
+            header_type = header['type']
+        headers_dicts.append(dict(id=header['id'], type=header_type))
 
     # Maintain data dictionaries from matching column names
     if existing_info: