Purpose: To investigate breast tissue compositions using a photon‐counting spectral Computed Tomography (CT) after spectral distortion correction. Methods: A total of 37 postmortem breast samples were scanned with a spectral CT based on a Cadmium‐Zinc‐Telluride (CZT) photon‐counting detector with five energy thresholds. All scans were acquired at beam energy of 100 kVp and a tube current of 1 mA. The energy thresholds of the CZT detector were set at 22, 42, 53, 65 and 90 keV. An image‐based spectral distortion correction method was used to eliminate the pulse pileup and other artifacts. Image‐based dual energy decomposition was performed with a three‐material calibration phantom to quantify postmortem breast tissue into water, lipid and protein contents. Results from dual energy decomposition were compared to chemical analysis. The Root Mean Square (RMS) errors for all samples were compared for images with and without spectral distortion correction. Results: The average RMS percentage errors of water, lipid, and protein over 37 breast samples, compared to data from chemical analysis, were 5.7% and 13.3% for images reconstructed with and without spectral distortion corrections, respectively. Conclusions: The results of this study suggest that the spectral distortion correction reduced the RMS percentage error of breast tissue composition by a factor greater than 2. With the proposed spectral distortion correction method, photon‐ counting detectors can be used for accurate quantitative tissue decomposition studies in spectral CT applications.