Dicentric chromosome analysis remains the most widely used method in biodosimetry. It has a lower detection limit of about 0.1 Gy, and allows one to distinguish between whole- and partial-body exposures. A drawback of the dicentric analysis is that it is a time consuming method and maybe difficult to implement in a mass casualty event. To try to increase the analysis capacity, automatic dicentric scoring (ADS) using image analysis software is being incorporated in several laboratories. Here we present the results obtained in an emergency exercise simulating 50 victims. The ability to distinguish different radiations scenarios is evaluated. To simulate whole-body exposures peripheral blood samples were irradiated at doses between 0–4.7 Gy, and to simulate partial-body exposures irradiated and nonirradiated blood were mixed in different proportions. With the data obtained from the first slide analyzed (with about 300–400 cells), 32 of 34 simulated whole-body exposures were correctly classified according to radiation exposure levels. For simulated partial-body irradiations, it was possible to detect them as partial exposures at the end of the first slide analyzed but only at the highest doses. In all cases the classification was updated every time the analysis of one additional slide was finished. The comparison between our present results and those reported in the literature for manual scoring shows that for triage purposes the ADS based on 300–400 cells is similar in efficiency to classifying the cases based on manual scoring of 50 cells. However, if one accounts for the associated uncertainties and the time needed for ADS, we suggest that ADS triage scoring should be based on about 1,000 cells. For final dose estimations the number of cells to score will depend on the initial estimated dose, and on the information contributed from physical dose-reconstruction or clinical symptoms. At doses higher than 1 Gy, we propose analysis of 1,500 and for lower doses or suspected partial-body exposures, the number of cells to score should be 3,000.