Left ventricle segmentation in 3D echocardiographic images has a crucial role in the analysis of cardiac structure and function and – more general – diagnosis of cardiovascular diseases. However, due to the relatively low image quality (i.e. contrast-to-noise ratio), low spatio-temporal resolution, and complexity of anatomical structures, this remains challenging. Recently, deep convolutional neural networks have been proposed for automatic segmentation in medical images, showing significant performance improvement over traditional segmentation algorithms in particular applications and might thus be able to offer a solution in 3D echocardiographic segmentation. However, annotated 3D echocardiography data as required to train deep learning models are not widely available. This paper introduces a new annotated 3D echocardiography dataset and aims to explore state-of-the-art deep learning models for volumetric left ventricle segmentation.