Purpose: Positron emission tomography (PET) of lung tumors suffers due to breathing‐motion induced blurring. Gated PET ameliorates motion blurring, but the concomitant reduction in coincidence‐event statistics per gated image leads to increased image noise, decreasing the utility of the images. In this work we demonstrate a method of reducing motion‐induced blurring from PET images without decreasing coincidence‐event statistics. Methods: The method works by deforming and stacking gated PET images using lung tissue motion vectors derived from deformable image registration performed between the phases of respiratory‐correlated computed tomography (4DCT) images. A duration of 12 minutes of respiratory‐correlated FDG‐PET data in list‐mode format and two 4DCT scans were acquired from 3 patients with a total of 4 small (4–18cc) lung tumors. A motion‐corrected image was produced using the first 2 minutes of list‐mode data. The motion‐corrected image was compared to a high‐statistics gated end‐exhale PET image created from the remaining 10 minutes of PET list‐mode data, which was used as a gold‐standard approximating a truly motion‐free image. The images were also compared against an un‐gated PET image produced using the first two minutes of list‐mode data, to quantify the improvement provided by the motion‐correction process. Results: Tumor SUVmax was on average 44% higher for motion‐reduced images relative to un‐gated images, but still on average 14% lower than the gold‐standard high‐statistics gated images. Volumes of tumors segmented at 40% of SUVmax were on average 32% lower relative to un‐gated images, but still 22% higher than the high‐statistics gated images. Conclusions: A process for reducing motion blurring in PET images without increasing image noise was demonstrated. The motion‐corrected images are quantitatively and qualitatively improved relative to un‐gated images, however further improvement is needed in order to match the quality of high‐statistics gated images.