diff --git a/pytorch_lightning/trainer/__init__.py b/pytorch_lightning/trainer/__init__.py
index 8cb144cb..554b34f8 100644
--- a/pytorch_lightning/trainer/__init__.py
+++ b/pytorch_lightning/trainer/__init__.py
@@ -30,6 +30,7 @@ This is the basic use of the trainer:
     trainer = Trainer()
     trainer.fit(model)
 
+
 --------
 
 Best Practices
@@ -59,6 +60,14 @@ So you can run it like so:distributed_backend
 
     $ python main.py --gpus 2
 
+
+.. note::
+    If you want to stop a training run early, you can press "Ctrl + C" on your keyboard.
+    The trainer will catch the `KeyboardInterrupt` and attempt a graceful shutdown, including
+    running callbacks such as `on_train_end`. The trainer object will also set an attribute
+    `interrupted` to `True` in such cases. If you have a callback which shuts down compute
+    resources, for example, you can conditionally run the shutdown logic for only uninterrupted runs.
+
 ------------
 
 Testing
diff --git a/pytorch_lightning/trainer/trainer.py b/pytorch_lightning/trainer/trainer.py
index dca3d539..7b0568e7 100644
--- a/pytorch_lightning/trainer/trainer.py
+++ b/pytorch_lightning/trainer/trainer.py
@@ -364,6 +364,7 @@ class Trainer(
         self.global_step = 0
         self.current_epoch = 0
         self.total_batches = 0
+        self.interrupted = False
 
         # configure logger
         self.configure_logger(logger)
diff --git a/pytorch_lightning/trainer/training_loop.py b/pytorch_lightning/trainer/training_loop.py
index 071a2f94..ba974dac 100644
--- a/pytorch_lightning/trainer/training_loop.py
+++ b/pytorch_lightning/trainer/training_loop.py
@@ -203,6 +203,7 @@ class TrainerTrainLoopMixin(ABC):
     use_amp: bool
     track_grad_norm: ...
     model: LightningModule
+    interrupted: bool
     running_loss: ...
     training_tqdm_dict: ...
     reduce_lr_on_plateau_scheduler: ...
@@ -387,6 +388,7 @@ class TrainerTrainLoopMixin(ABC):
 
         except KeyboardInterrupt:
             log.info('Detected KeyboardInterrupt, attempting graceful shutdown...')
+            self.interrupted = True
             self.run_training_teardown()
 
     def run_training_epoch(self):
diff --git a/tests/trainer/test_trainer.py b/tests/trainer/test_trainer.py
index ddf79cb2..f80870ec 100644
--- a/tests/trainer/test_trainer.py
+++ b/tests/trainer/test_trainer.py
@@ -12,6 +12,7 @@ from pytorch_lightning.callbacks import (
     EarlyStopping,
     ModelCheckpoint,
 )
+from pytorch_lightning import Callback
 from pytorch_lightning.core.lightning import load_hparams_from_tags_csv
 from pytorch_lightning.trainer.logging import TrainerLoggingMixin
 from pytorch_lightning.utilities.exceptions import MisconfigurationException
@@ -630,3 +631,33 @@ def test_nan_params_detection(tmpdir):
     # after aborting the training loop, model still has nan-valued params
     params = torch.cat([param.view(-1) for param in model.parameters()])
     assert not torch.isfinite(params).all()
+
+
+def test_trainer_interrupted_flag(tmpdir):
+    """Test the flag denoting that a user interrupted training."""
+
+    model = DictHparamsModel({'in_features': 28 * 28, 'out_features': 10})
+
+    class InterruptCallback(Callback):
+        def __init__(self):
+            super().__init__()
+
+        def on_batch_start(self, trainer, pl_module):
+            raise KeyboardInterrupt
+
+    interrupt_callback = InterruptCallback()
+
+    trainer_options = {
+        'callbacks': [interrupt_callback],
+        'max_epochs': 1,
+        'val_percent_check': 0.1,
+        'train_percent_check': 0.2,
+        'progress_bar_refresh_rate': 0,
+        'logger': False,
+        'default_save_path': tmpdir,
+    }
+
+    trainer = Trainer(**trainer_options)
+    assert not trainer.interrupted
+    trainer.fit(model)
+    assert trainer.interrupted