Purpose-Respiratory motion is a significant source of anatomic uncertainty in radiotherapy treatment planning, and may result in errors of portal size and subsequent radiation dose. While 4DCT allows for more accurate analysis of the respiratory cycle, breathing irregularities during data acquisition may cause considerable image distortions. The aim of this study was to examine the impact of respiratory irregularities on 4DCT imaging, and to evaluate a novel image reconstruction algorithm based upon percentile-based tagging of the respiratory cycle.Materials/Methods-Respiratory correlated helical CT was acquired on 11 consecutive patients. Inspiration and expiration datasets were reconstructed using the default phase based method as well as a novel respiration percentile based method using patient specific metrics to define ranges of the reconstruction. Image output was analyzed in a blinded fashion for both the phase and percentile based reconstructions to determine the prevalence and severity of image artifacts.Results-The percentile-based algorithm resulted in a significant reduction in artifact severity compared to the phase-based algorithm, although overall artifact prevalence did not differ between the two algorithms. The magnitude of differences in respiratory tag placement between the phase and percentile based algorithms was correlated with the presence of image artifacts.Conclusions-We conclude that our novel 4DCT reconstruction method may be useful to detect clinically relevant image distortions that may otherwise go unnoticed, and to reduce image distortion associated with some respiratory irregularities. Further work is necessary to assess the clinical impact on areas of possible irregular breathing.