diff --git a/cloudkitty/writer/csv_map.py b/cloudkitty/writer/csv_map.py
index 5c1803e0..f0d3e8a9 100644
--- a/cloudkitty/writer/csv_map.py
+++ b/cloudkitty/writer/csv_map.py
@@ -95,7 +95,8 @@ class CSVMapped(csv_base.BaseCSVBackend):
         """Get the end usage of this period.
 
         """
-        if self.cached_start == self.usage_start and self.cached_end_str:
+        if self.cached_start == self.usage_start and self.cached_end_str \
+           and self.cached_end > self.cached_start:
             return self.cached_end_str
         else:
             usage_end = self.usage_start_dt + datetime.timedelta(
diff --git a/releasenotes/notes/fix-csv-usage-end-7bcf4cb5effc4461.yaml b/releasenotes/notes/fix-csv-usage-end-7bcf4cb5effc4461.yaml
new file mode 100644
index 00000000..8137d8a5
--- /dev/null
+++ b/releasenotes/notes/fix-csv-usage-end-7bcf4cb5effc4461.yaml
@@ -0,0 +1,4 @@
+---
+fixes:
+  - |
+    The value of the UsageEnd field in CSV reports has been fixed.