The purpose of this work was to increase the resolution of hyperpolarized metabolic imaging of the rat heart with accelerated imaging using k–t principal component analysis (k–t PCA) and k–t compressed sensing (k–t SPARSE). Fully sampled in vivo datasets were acquired from six healthy rats after the injection of hyperpolarized [1‐13C]pyruvate. Data were retrospectively undersampled and reconstructed with either k–t PCA or k–t SPARSE. Errors of signal–time curves of pyruvate, lactate and bicarbonate were determined to compare the two reconstruction algorithms for different undersampling factors R. Prospectively undersampled imaging at 1 × 1 × 3.5‐mm3 resolution was performed with both methods in the same animals and compared with the fully sampled acquisition. k–t SPARSE was found to perform better at R < 3, but was outperformed by k–t PCA at R ≥ 4. Prospectively undersampled data were successfully reconstructed with both k–t PCA and k–t SPARSE in all subjects. No significant difference between the undersampled and fully sampled data was found in terms of signal‐to‐noise ratio (SNR) performance and metabolic quantification. Accelerated imaging with both k–t PCA and k–t SPARSE allows an increase in resolution, thereby reducing the intravoxel dephasing of hyperpolarized metabolic imaging of the rat heart.