Cardiac strains are important biomarkers of cardiac performance. Regional strain calculation from Cine MR images would facilitate their clinical adoption and obviate the need for acquiring specialized images. In this work we investigate three models to compute mid-ventricular cardiac strains. The first model is based on two short axis DENSE MRI slices, which provide voxelwise displacements, i.e., a complete description of the myocardial deformation. The second model is based on one DENSE slice, where we investigate the effect of reduced displacement information – but faster data acquisition – on the computed strains. The third model is based on one short-axis Cine slice, which is among the fastest and most common MRI data acquired in cardiac exams. To validate the three models and the underlying kinematic assumptions, we use a simple and previously developed computational cardiac phantom. Subsequently, we compare the proposed models using data from healthy volunteers (N = 10), where we consider the results from the two-slice DENSE model as the most reliable since they are computed from the complete voxelwise displacement field. All models are evaluated using circumferential and longitudinal strains. Although all models agree well with the ground truth analytical solution when images are generated from the phantom, discrepancies emerge as the models are applied to real data: the one-slice DENSE model tends to significantly underestimate longitudinal strains while the one-slice Cine model overall underestimates epicardial circumferential strains. These differences are discussed in terms of the models’ characteristics together with possible future improvements.