An MR method determines local relaxation time values (T1, T2) of an examination object (5). First, two or more MR images (3, 4) of the examination object (5) are recorded, each with different time parameter sets (TR1, TE1, TR2, TE2) of an imaging sequence. MR images (6, 7) of a phantom are likewise recorded using the same time parameter sets (TR1, TE1, TR2, TE2) of the imaging sequence. The phantom has a known spatial distribution of relaxation time values (T1, T2). The local relaxation time values (T1, T2) of the examination object (5) are determined by comparing image values of the MR images (3, 4) of the examination object (5) with image values of the MR images (6, 7) of the phantom and by assigning image values of the MR images (6, 7) of the phantom to relaxation time values (T1, T2) according to the known spatial distribution of relaxation time values (T1, T2) of the phantom.