diff --git a/doc/acas_xu.rst b/doc/acas_xu.rst
index eb58bc132fca57cf1275d1a6785cfb4818263f7d..e88ac4db16f7c13c144b8f48ba1c26ec63665116 100644
--- a/doc/acas_xu.rst
+++ b/doc/acas_xu.rst
@@ -398,9 +398,6 @@ sub-results to provide the final result corresponding to the original goal
 formula.
 
 
-References
-==========
-
 .. [Manfredi2016] G. Manfredi and Y. Jestin, *An introduction to ACAS Xu and the
    challenges ahead*, 2016 IEEE/AIAA 35th Digital Avionics Systems Conference
    (DASC), 2016, pp. 1-9, doi: 10.1109/DASC.2016.7778055
diff --git a/doc/mnist.rst b/doc/mnist.rst
index 0d9c57ed8f243095ce2824df43691967c2fc6af3..c80180c615e20f6cf9aeed8fa74a23ae3ac00930 100644
--- a/doc/mnist.rst
+++ b/doc/mnist.rst
@@ -152,8 +152,6 @@ for which nnenum is sure that the model ``MNIST_256_2.onnx`` is not robust with
 respect to :math:`1 \%` perturbation. At the moment, CAISAR is not able to tell
 which are the images in the dataset that cause such result.
 
-References
-==========
 
 .. [LiDeng2012] Li Deng, *The MNIST Database of Handwritten Digit Images for
                Machine Learning Research*, IEEE Signal Process. Mag., 2012, pp.