PurposeRadiotherapy using grids containing cm‐wide beam elements has been carried out sporadically for more than a century. During the past two decades, preclinical research on radiotherapy with grids containing small beam elements, 25 μm–0.7 mm wide, has been performed. Grid therapy with larger beam elements is technically easier to implement, but the normal tissue tolerance to the treatment is decreasing. In this work, a new approach in grid therapy, based on irradiations with grids containing narrow carbon‐ion beam elements was evaluated dosimetrically. The aim formulated for the suggested treatment was to obtain a uniform target dose combined with well‐defined grids in the irradiated normal tissue. The gain, obtained by crossfiring the carbon‐ion beam grids over a simulated target volume, was quantitatively evaluated.MethodsThe dose distributions produced by narrow rectangular carbon‐ion beams in a water phantom were simulated with the PHITS Monte Carlo code. The beam‐element height was set to 2.0 cm in the simulations, while the widths varied from 0.5 to 10.0 mm. A spread‐out Bragg peak (SOBP) was then created for each beam element in the grid, to cover the target volume with dose in the depth direction. The dose distributions produced by the beam‐grid irradiations were thereafter constructed by adding the dose profiles simulated for single beam elements. The variation of the valley‐to‐peak dose ratio (VPDR) with depth in water was thereafter evaluated. The separation of the beam elements inside the grids were determined for different irradiation geometries with a selection criterion.ResultsThe simulated carbon‐ion beams remained narrow down to the depths of the Bragg peaks. With the formulated selection criterion, a beam‐element separation which was close to the beam‐element width was found optimal for grids containing 3.0‐mm‐wide beam elements, while a separation which was considerably larger than the beam‐element width was found advantageous for grids containing 0.5‐mm‐wide beam elements. With the single‐grid irradiation setup, the VPDRs were close to 1.0 already at a distance of several cm from the target. The valley doses given to the normal tissue at 0.5 cm distance from the target volume could be limited to less than 10% of the mean target dose if a crossfiring setup with four interlaced grids was used.ConclusionsThe dose distributions produced by grids containing 0.5‐ and 3.0‐mm wide beam elements had characteristics which could be useful for grid therapy. Grids containing mm‐wide carbon‐ion beam elements could be advantageous due to the technical ease with which these beams can be produced and delivered, despite the reduced threshold doses observed for early and late responding normal tissue for beams of millimeter width, compared to submillimetric beams. The treatment simulations showed that nearly homogeneous dose distributions could be created inside the target volumes, combined with low valley doses in the normal tissue located close to the target volume, if the carbon‐ion beam grids were crossfired in an interlaced manner with optimally selected beam‐element separations. The formulated selection criterion was found useful for the quantitative evaluation of the dose distributions produced by the different irradiation setups.