diff --git a/.circleci/config.yml b/.circleci/config.yml
new file mode 100644
index 0000000..1033fb0
--- /dev/null
+++ b/.circleci/config.yml
@@ -0,0 +1,23 @@
+version: 2.1
+
+jobs:
+  test:
+    parameters:
+      python-version:
+        type: string
+    docker:
+      - image: circleci/python:<< parameters.python-version >>
+    steps:
+      - checkout
+      - run: pip3 install --upgrade pip
+      - run: pip3 install -r requirements.txt
+      - run: pip3 install pytest
+      - run: pytest
+
+workflows:
+  all-tests:
+    jobs:
+      - test:
+          matrix:
+            parameters:
+              python-version: ["3.5", "3.6", "3.7", "3.8", "3.9"]
diff --git a/README.md b/README.md
index a0919d1..6cef259 100644
--- a/README.md
+++ b/README.md
@@ -1,6 +1,6 @@
 Extract the MODS metadata of a bunch of METS files into a pandas DataFrame.
 
-[![Build Status](https://travis-ci.org/qurator-spk/modstool.svg?branch=master)](https://travis-ci.org/qurator-spk/modstool)
+[![Build Status](https://circleci.com/gh/qurator-spk/modstool.svg?style=svg)](https://circleci.com/gh/qurator-spk/modstool)
 
 Column names are derived from the corresponding MODS elements. Some domain
 knowledge is used to convert elements to a useful column, e.g. produce sets