diff --git a/README.rst b/README.rst
index b9acc3edc..f526f2727 100644
--- a/README.rst
+++ b/README.rst
@@ -307,6 +307,21 @@ Processing each frame in a framed message involves a certain amount of overhead.
 increasing the frame size can offer potentially significant performance gains.  We recommend that you tune these values
 to your use-case in order to obtain peak performance.
 
+Thread safety
+==========================
+The ``EncryptionSDKClient`` and all provided ``CryptoMaterialsManager`` are thread safe.
+But instances of ``BaseKMSMasterKeyProvider`` MUST not be shared between threads,
+for the reasons outlined in `the boto3 docs <https://boto3.amazonaws.com/v1/documentation/api/latest/guide/resources.html#multithreading-or-multiprocessing-with-resources>`_.
+
+Because the ``BaseKMSMaterKeyProvider`` creates a `new boto3 sessions <https://github.com/aws/aws-encryption-sdk-python/blob/08f305a9b7b5fc897d9cafac55fb98f3f2a6fe13/src/aws_encryption_sdk/key_providers/kms.py#L665-L674>`_ per region,
+users do not need to create a client for every region in every thread;
+a new  ``BaseKMSMasterKeyProvider`` per thread is sufficient.
+
+(The ``BaseKMSMasterKeyProvider`` is the internal parent class of all the KMS Providers.)
+
+Finally, while the ``CryptoMaterialsCache`` is thread safe,
+sharing entries in that cache across threads needs to be done carefully
+(see the !Note about partition name `in the API Docs <https://aws-encryption-sdk-python.readthedocs.io/en/latest/generated/aws_encryption_sdk.materials_managers.caching.html#aws_encryption_sdk.materials_managers.caching.CachingCryptoMaterialsManager>`_).
 
 .. _AWS Encryption SDK: https://docs.aws.amazon.com/encryption-sdk/latest/developer-guide/introduction.html
 .. _cryptography: https://cryptography.io/en/latest/