Purpose: To analyze toxicity of hypofractionated lung SBRT according to QUANTEC recommendations.Materials and Methods: 44 patients with small primary lung cancer or oligometastasis were treated with SBRT.Patients were immobilized using SBRT frame.We exported composite dose volume histograms (DVHs) of gross target volume (GTV), planning target volume (PTV) and normal lung tissue (whole lung-GTV).Normal tissue complication probability (NTCP) based on Poisson model was calculated using normal lung DVH, and normalized total dose (NTD) volume histograms were generated at 2 Gy fractions with / =3.7 Gy to account for different fractionations.Variable importance for projection (VIP) and correlation coefficient were calculated using partial least squares regression (PLSR) and logistic regression (LR) to identify prediction factors for normal tissue complications.Results: There were 2 (4.2%) local failures, 4 (9.1%) grade 2 pneumonitis and 3 (6.8%) radiation-induced fibrosis.The <10% pneumonitis rate is in line with QUANTEC report for lung SBRT.5/14 (35.7%) of lesions in fight lower lobe developed complications, in comparison to 1/16 (6.3%) in fight upper and middle lobes, 0/5 (0%) in left lower lobe and 1/11 (9.1%) in left upper lobe.The most influential VIP for predicting complications were mean GTV dose (VIP =1.500, R=0.145) and GTV volume (VIP=1.451, R=-0.140), followed by NTCP (VIP=1.106, R=0.107) although calculated probability was much higher than frequency of complications.LR was able to produce a reasonable NTCP curve for these variables.The predictions factors used for conventional lung RT complications, i.e., V20, VNTD20, mean dose and mean NTD of normal lung had moderate VIP, were negatively correlated to a NTCP logistic curve.Conclusions: Radiation induced complication for SBRT increases with mean GTV dose and cannot be predicted with known prediction factors for conventional lung RT.