Single‐shot echo planar imaging (EPI), which allows an image to be acquired using a single excitation pulse, is used widely for imaging the metabolism of hyperpolarized 13C‐labelled metabolites in vivo as the technique is rapid and minimizes the depletion of the hyperpolarized signal. However, EPI suffers from Nyquist ghosting, which normally is corrected for by acquiring a reference scan. In a dynamic acquisition of a series of images, this results in the sacrifice of a time point if the reference scan involves a full readout train with no phase encoding. This time penalty is negligible if an integrated navigator echo is used, but at the cost of a lower signal‐to‐noise ratio (SNR) as a result of prolonged T
2* decay. We describe here a workflow for hyperpolarized 13C EPI that requires no reference scan. This involves the selection of a ghost‐containing background from a 13C image of a single metabolite at a single time point, the identification of phase correction coefficients that minimize signal in the selected area, and the application of these coefficients to images acquired at all time points and from all metabolites. The workflow was compared in phantom experiments with phase correction using a 13C reference scan, and yielded similar results in situations with a regular field of view (FOV), a restricted FOV and where there were multiple signal sources. When compared with alternative phase correction methods, the workflow showed an SNR benefit relative to integrated 13C reference echoes (>15%) or better ghost removal relative to a 1H reference scan. The residual ghosting in a slightly de‐shimmed B
0 field was 1.6% using the proposed workflow and 3.8% using a 1H reference scan. The workflow was implemented with a series of dynamically acquired hyperpolarized [1‐13C]pyruvate and [1‐13C]lactate images in vivo, resulting in images with no observable ghosting and which were quantitatively similar to images corrected using a 13C reference scan.