In stereotactic body radiotherapy (SBRT), the respiratory tumor motion makes target definition very important to achieve optimal clinical results for treatment of early stage lung cancer. In this article, the authors quantitatively evaluated the influence of different target definition strategies on image-guided respiratory-gated SBRT for lung cancer. Twelve lung cancer patients with 4D CT estimated target motion of >1 cm were selected for this retrospective study. An experienced physician contoured gross target volumes (GTVs) at each 4D CT phase for all patients. Three types of internal target volumes (ITVs) were generated based on the contoured GTVs:(1) ITVBH: GTV contoured on deep expiration breath-hold (BH) CT with an isotropic internal margin (IM) of 5 mm; (2) ITV50: GTV contoured at the end-expiration (50%) phase with an isotropic IM of 5 mm; (3) ITVGW: Composite volume of all GTVs within the gating window, defined as several phases around phase 50% with residual target motion of <5 mm. Planning target volumes (PTVs) were generated by adding 3 mm isotropic setup error margin to ITVs. Three treatment plans, namely, PlanBH, Plan50, and PlanGW, were created based on the three PTVs. Identical beam settings and planning constraints were used for all three plans for each patient. The prescription dose was 60 Gy in three fractions. The potential toxicities to the critical organs were quantified by mean lung dose (MLD), lung volume receiving >20 Gy (V20), mean heart dose (MHD), and spinal cord dose (SCD). It is shown that the tumor volume and dose coverage are comparable for PlanBH and Plan50. On average, PTVGW are 38% less than PTV50. Although for most patients PTV50 encompasses the entire PTVGW, up to 5.48 cm3 (6%) of PTVGW is outside PTV50. Compared to Plan50, prescribed percentage is about 2% higher for PlanGW, and the average dose decreases in critical organs are 0.78 Gy for MLD, 1.02% for V20, 0.61 Gy for MHD and 0.59 Gy for maximum SCD. For the cases receiving high lung and heart dose with Plan50, the dose reduction is 1.0 Gy for MLD and 1.14 Gy for MHD with PlanGW. Our preliminary results show that a patient-specific ITV, defined as the composite volume of all GTVs within the gating window, may be used to define PTV in image-guided respiratory-gated SBRT. This approach potentially reduces the irradiated volume of normal tissue further without sacrificing target dose coverage and thus may minimize the risk of treatment-related toxicities.