Purpose: Rotational errors in patient setup are well characterized with cone beam CT imaging. One typical approach makes 3D and 4D corrections only, and continues with treatment. Due to the difficulty of computing the effect of rotations on delivered dose, many centers utilize class solutions based on treatment site to determine if roll and pitch can generally be neglected. This study examines the effects of rotation on dose distribution for individuals and classes. Methods: Proprietary software was developed that automates the entire evaluation process which begins with DICOM files exported from Eclipse 8.6. The effects of patient rotation perturbations on the dose‐volume histogram are compared to the baseline DVH via TCP‐NTCP analysis. A retrospective study was conducted using eight nasal tumor canine patients. Virtual rotation perturbations of up to 10 degrees about a vector set uniformly spaced over a 2π hemisphere were induced and doses to target and organs were recomputed. For this study, the evaluation was limited to a single organ proximal to the tumor (left or right eye lens). For each patient, the changes to the original baseline values (dTCP and dNTCP) were computed as a function of rotational error. Individual and group metrics were obtained. Results: At 10 degrees rotation mean dTCP decreases by 53% (S.D. +/−66%), while mean dNTCP values increase 6% (S.D. +/−22%). Both begin to make sharp deviations from their base values around 3 to 4 degrees deviation, however statistical analysis shows no significant deviation from the baseline out to 10 degrees (p=0.05). For individual cases dTCP reductions of 50% as early as 4 degrees rotation were observed. Conclusion: While class solutions to errors in rotation are helpful for general solutions, a patient by patient analysis for dose changes with rotational errors may be required to avoid significant deviations between planned and delivered dose.