We propose a k-domain spline interpolation method with constrained polynomial fit based on spectral phase in swept-source optical coherence tomography (SS-OCT). A Mach–Zehnder interferometer (MZI) unit is connected to the swept-source of the SS-OCT system to generate calibration signal in sync with the fetching of interference spectra. The spectral phase of the calibration signal is extracted by Hilbert transformation. The fitted phase–time relationship is obtained by polynomial fitting with the constraint of passing through the central spectral phase. The fitting curve is then adopted for k-domain uniform interpolation based on evenly spaced phase. In comparison with conventional k-domain spline interpolation, the proposed method leads to improved axial resolution and peak response of the axial point spread function (PSF) of the SS-OCT system. Enhanced performance resulting from the proposed method is further verified by OCT imaging of a home-constructed microspheres-agar sample and a fresh lemon. Besides SS-OCT, the proposed method is believed to be applicable to spectral domain OCT as well.