Epilepsy is a chronic neurological disorder. Epileptics are prone to sudden seizures that cause disruptions in their daily lives. The separation of epileptic and non-epileptic activity on the electroencephalogram (EEG) and identification of the form of epileptic activity play critical roles in providing patients with appropriate treatment. To recognize epileptic seizures, medical experts visually inspect recordings of EEG signals, which require much time and effort. Therefore, a seizure detection system can improve the monitoring and diagnosis of epilepsy and reduce the doctors' workload. This paper presents an end-to-end automated seizure detection method based on deep learning that does not require considerable EEG data preprocessing or feature extraction. As a result presents a one-dimensional convolutional neural network-long short-term memory (1D-CNN-LSTM) model for differentiating normal, ictal, and interictal EEG data. This method is evaluated using the University of Bonn (BoU) and Neurology and Sleep Centre database (NSC). We achieved accuracy values of 99–100% for the BoU dataset and 100% for the NSC dataset with our best model. In contrast with recent studies, our hybrid automated approach does not require any pre-selected features to be estimated and shows high performance with promising possibilities for their use in clinical practice.