High-fidelity fault detection on seismic images is one of the most important and challenging topics in the field of automatic seismic interpretation. Conventional hand-picking-based and semi-human-intervened fault-detection approaches are being replaced by fully automatic methods thanks to the development of machine learning. We have developed a novel multiscale attention convolutional neural network (MACNN) to improve machine-learning-based automatic end-to-end fault detection on seismic images. The most important characteristics of our MACNN fault-detection method are that it uses a multiscale spatial-channel attention mechanism to merge and refine encoder feature maps of different spatial resolutions. The new architecture enables our MACNN to more effectively learn and exploit contextual information embedded in the encoder feature maps. We determine through several synthetic data and field data examples that our MACNN tends to produce higher resolution, higher fidelity fault maps from complex seismic images compared to those of the conventional fault-detection convolutional neural network, thus leading to improved geologic fidelity and interpretability of detected faults.