This paper sets out guidelines for managing radiation exposure incidents involving patients in diagnostic and interventional radiology. The work is based on collation of experiences from representatives of international and national organizations for radiologists, medical physicists, radiographers, regulators, and equipment manufacturers, derived from an International Atomic Energy Agency Technical Meeting. More serious overexposures can result in skin doses high enough to produce tissue reactions, in interventional procedures and computed tomography, most notably from perfusion studies. A major factor involved has been deficiencies in training of staff in operation of equipment and optimization techniques. The use of checklists and time outs before procedures commence, and dose alerts when critical levels are reached during procedures can provide safeguards to reduce risks of these effects occurring. However, unintended and accidental overexposures resulting in relatively small additional doses can take place in any diagnostic or interventional X-ray procedure and it is important to learn from errors that occur, as these may lead to increased risks of stochastic effects. Such events may involve the wrong examinations, procedural errors, or equipment faults. Guidance is given on prevention, investigation and dose calculation for radiology exposure incidents within healthcare facilities. Responsibilities should be clearly set out in formal policies, and procedures should be in place to ensure that root causes are identified and deficiencies addressed. When an overexposure of a patient or an unintended exposure of a foetus occurs, the foetal, organ, skin and/or effective dose may be estimated from exposure data. When doses are very low, generic values for the examination may be sufficient, but a full assessment of doses to all exposed organs and tissues may sometimes be required. The use of general terminology to describe risks from stochastic effects is recommended rather than calculation of numerical values, as these are misleading when applied to individuals.