In this study, we reported a complex-signal-based optical coherence tomography angiography (OCTA) method, called complex correlated phase gradient variance (CCPGV), for mapping high-quality microvascular images. The performance of the newly proposed algorithm is benchmarked against the previously reported phase-resolved Doppler variance, complex differential variance (CDV), and split-spectrum amplitude and phase-gradient angiography (SSAPGA), by both tissue phantoms experiments and in vivo human skin measurements. Compared to the phase-resolved Doppler variance, CCPGV can intrinsically reject undesirable phase shifts caused by bulk motion and trigger jitter. In contrast to CDV and SSAPGA, which are insensitive to phase instability, CCPGV can provide superior motion contrast, as demonstrated by ∼1.4 and 3 times higher contrast in phantom experiments respectively. An increase of 27.3% and 106.3% were found in vivo experiments for CCPGV, making it easier to distinguish vessels from the background static. Benefiting from the advantages, more vessels and better connectivity can be visualized in the en-face angiogram processed by the CCPGV method. We believe that our method will benefit the biomedical community to some extent in disease diagnosis and monitoring.