The Section of Forensic Medicine at the University of Copenhagen performs age evaluations of the living at the request of the police. The established procedure in these cases involves: (1) a physical examination; (2) an odontological examination based on evaluation of an orthopantomogram and intraoral dental radiographs. Different methods are used depending on the maturity of the individual examined; and (3) a carpal X-ray examination, using the Greulich and Pyle Atlas (GPA) method. We present the results of intra- and interobserver tests of carpal X-rays in blind trials, and a comparison of the age estimations by carpal X-rays and odontological age estimation. We retrieved 159 cases from the years 2000-2002 (inclusive). The intra- and interobserver errors are overall small. We found full agreement in 126/159 cases, and this was between experienced users and novices. Overall, the mean difference was 0.053 years, with a standard deviation of +/-0.567 years. Our results show that the direct and naïve use of the GPA is simple and reproducible, even when applied by non-experienced users. This is probably also why the method is still widely applied, even though other authors have stated that atlas-based techniques are obsolete and ought to be replaced by other methods. Specifically, the GPA test sample consisted of American "white" children "above average in economic and educational status", leading to the question as to how comparable subjects being scored by the GPA method today are to this original sample. Indeed, we found a minor difference in the age estimations for older juveniles between odontologists and radiologists, which may indicate that more studies on comparative dental and skeletal aging for geographically different populations would be beneficial. Finally, if using the GPA method in legal cases, we would advise that one should closely adhere to the standard deviations given, and intermediary scores should be noted and not just a resultant age. The final age statement should also address the problems of reference populations.