diff --git a/CHANGELOG.rst b/CHANGELOG.rst
index a7757cad0f..4977dbad9b 100644
--- a/CHANGELOG.rst
+++ b/CHANGELOG.rst
@@ -8,6 +8,7 @@ CHANGELOG
 * doc-fix: Remove incorrect parameter for EI TFS Python README
 * feature: ``Predictor``: delete SageMaker model
 * feature: ``Pipeline``: delete SageMaker model
+* bug-fix: Estimator.attach works with training jobs without hyperparameters
 
 1.18.3.post1
 ============
diff --git a/src/sagemaker/estimator.py b/src/sagemaker/estimator.py
index a0af44c2d7..f840ce17fc 100644
--- a/src/sagemaker/estimator.py
+++ b/src/sagemaker/estimator.py
@@ -425,7 +425,9 @@ def _prepare_init_params_from_job_description(cls, job_details, model_channel_na
         init_params['output_path'] = job_details['OutputDataConfig']['S3OutputPath']
         init_params['output_kms_key'] = job_details['OutputDataConfig']['KmsKeyId']
 
-        init_params['hyperparameters'] = job_details['HyperParameters']
+        has_hps = 'HyperParameters' in job_details
+        init_params['hyperparameters'] = job_details['HyperParameters'] if has_hps else {}
+
         if 'TrainingImage' in job_details['AlgorithmSpecification']:
             init_params['image'] = job_details['AlgorithmSpecification']['TrainingImage']
         elif 'AlgorithmName' in job_details['AlgorithmSpecification']:
diff --git a/tests/unit/test_estimator.py b/tests/unit/test_estimator.py
index 59d803254e..e9c483b9b4 100644
--- a/tests/unit/test_estimator.py
+++ b/tests/unit/test_estimator.py
@@ -451,6 +451,20 @@ def test_attach_framework(sagemaker_session):
     assert framework_estimator.encrypt_inter_container_traffic is False
 
 
+def test_attach_without_hyperparameters(sagemaker_session):
+    returned_job_description = RETURNED_JOB_DESCRIPTION.copy()
+    del returned_job_description['HyperParameters']
+
+    mock_describe_training_job = Mock(name='describe_training_job',
+                                      return_value=returned_job_description)
+    sagemaker_session.sagemaker_client.describe_training_job = mock_describe_training_job
+
+    estimator = Estimator.attach(training_job_name='job',
+                                 sagemaker_session=sagemaker_session)
+
+    assert estimator.hyperparameters() == {}
+
+
 def test_attach_framework_with_tuning(sagemaker_session):
     returned_job_description = RETURNED_JOB_DESCRIPTION.copy()
     returned_job_description['HyperParameters']['_tuning_objective_metric'] = 'Validation-accuracy'