Objective: Most radiotherapy patients with prostate cancer are treated with volumetric modulated arc therapy (VMAT). Advantages of VMAT may be limited by daily treatment uncertainties such as setup errors, internal organ motion, and deformation. The position and shape of prostate target as well as normal organ, i.e., rectum volume around the target, may change during the course of treatment. The aim of the present work is to estimate rectal toxicity estimation using a novel two-level biological knowledge-based fuzzy logic method. Both prostate and rectal internal motions as well as setup uncertainties are considered without compromising target dose distribution in the present study. Materials and Methods: The Mamdani-type fuzzy logic framework was considered in two levels. The prostate target volume changes from minimum to maximum during the course of treatment. In the first level, the fuzzy logic was applied for determining biological acceptable target margin using tumor control probability and normal tissue complication probability (NTCP) parameters based on prostate target motion limits, and then, fuzzy margin was derived. The output margin of first-level fuzzy logic was compared to currently used margins. In second-level fuzzy, rectal volume variation with weekly analysis of cone-beam computed tomography (CBCT) was considered. The biological parameter (NTCP) was calculated corresponding to rectal subvolume variation with weekly CBCT image analysis. Using irradiated volume versus organ risk relationship from treatment planning, the overlapped risk volumes were estimated. Fuzzy rules and membership function were used based on setup errors, asymmetrical nature of organ motion, and limitations of normal tissue toxicity in Mamdani-type Fuzzy Inference System. Results: For total displacement, standard errors of prostate ranging from 0 to 5 mm range were considered in the present study. In the first level, fuzzy planning target volume (PTV) margin was found to be similar or up to 0.5 mm bigger than the conventional margin, but taking the modeling uncertainty into account resulted in a good match between the calculated fuzzy PTV margin and conventional margin formulations under error 0-5 mm standard deviation (SD) range. With application of fuzzy margin obtained from first-level fuzzy, overlapped rectal volumes and corresponding NTCP values were fuzzified in second-level fuzzy using rectal volume variations. The final risk factor (RF) of rectum was qualitatively assessed and found clinically acceptable for each fractional volume of irradiated to total volume and relevant NTCP values. The reason may be at 5 mm SD displacement error range, NTCP values would be within acceptable limit without compromising the tumor dose distribution though the confounding factors such as organ motion, deformation of rectum, and in-house image matching protocols exist. Conclusion: A new approach of two-level fuzzy logic may be suitable to estimate possible organ-at-risk (OAR) toxicity biologically without compromising tumor volume that includes both prostate target and OAR rectum deformation even at displacement standard errors of prostate ranging from 0 to 5 mm range which was considered in the present study. Using proposed simple and fast method, there is an interplay between volume-risk relationship and NTCP of OARs to judge real-time normal organ risk level or alter the treatment margins, particularly concern to individual factors such as comorbidities, genetic predisposition, and other lifestyle choices even at high displacement errors >5 mm SD range. [ABSTRACT FROM AUTHOR]