Objective: The published literature maintains that distortion correction of digital subtraction angiography is essential for accurate definition of the arteriovenous malformation (AVM) nidus when performing Gamma Knife radiosurgery. However, as Gamma Knife treatments usually involve small volume targets, which are purposely positioned as close to the centre of the frame as possible, we postulated that the angiographic target images should be minimally distorted. Thus, the use of uncorrected images would not significantly compromise the accuracy of the treatment. We investigated this postulate. Design: Patient and phantom images with and without distortion correction were assessed. The errors measured in the angiograms were compared to those of MR and CT images of the same subjects. Any errors less than 0.5 mm in any modality were accepted as intrinsic. Results: All errors measured in the study were sub-millimetre. In the patient group, errors with distortion correction averaged 0.13 mm (range 0– 0.3 mm). Without distortion correction, errors averaged 0.34 mm (range 0.1–0.6 mm). By comparison, the average error for intracranial MR imaging was 0.3 mm, with a maximum of 0.6 mm. Conclusion: For small, intracranial AVMs distortion correction did not significantly reduce the errors associated with the definition of the stereotactic co-ordinates. We suggest that prior to the purchase of this software, similar in-house testing should be performed.