In a method and magnetic resonance (MR) system to determine an MR relaxation time (for example a T1, T2 or T2* relaxation time) in the heart muscle in a magnetic resonance examination, a determination an annular slice image region of the heart muscle of the left heart chamber in MR image data with the use of an automatic image segmentation. Multiple sub-regions within the slice image region are automatically determined. Each sub-region respectively includes multiple pixels of the annular slice image region of the heart muscle of the left heart chamber. An MR relaxation time is determined automatically for each of the multiple sub-regions and associated with the corresponding sub-region. A characteristic MR relaxation time in the heart muscle is determined by a statistical analysis of the multiple MR relaxation times that are associated with the multiple sub-regions.