The human brain, one of the most vital organs, is responsible for controlling and coordinating the functions of the different organs in body systems. Sadly, brain tumors are a grave danger to this complex organ, frequently resulting in major disruptions to its operation. To ensure effective treatment and improve patient outcomes, the early and accurate detection of abnormal cells has become of utmost importance for one of the deadliest diseases. To address this, we present a robust deep learning model called 3D Res-UNet for automatic segmentation of 3D multimodal brain MRI. Using a 3D evolutionary U-shaped network that is based on changing the classical convolution block of the U-Net architecture to residual block, we use this model in our experiment. This model achieves high accuracy, precision, recall, and Dice scores for segmenting multiple tumor regions. The BraTS 2020 dataset has been used for training and evaluation, with encouraging outcomes, indicating its high application potential to assist clinical diagnosis and treatment decisions.