Purpose: Micrometer spatial resolution dosimetry has become inevitable for advanced radiotherapy techniques. A new approach using radiochromic films was developed to measure a radiation dose at a micrometer spatial resolution by confocal Raman spectroscopy. Methods: The commercial radiochromic films (RCF), EBT3 and EBT-XD, were irradiated with known doses using 50, 100, 200, and 300 kVp, and 6-MV x rays. The dose levels ranged from 0.3 to 50 Gy. The Raman mapping technique developed in our early study was used to readout an area of 100 9 100 µm 2 on RCF with improved lateral and depth resolutions with confocal Raman spectrometry. The variation in Raman spectra of C-C-C deformation and CC stretching modes of diacetylene polymers around 676 and 2060 cm À1 , respectively, as a function of therapeutic x-ray doses, was measured. The single peak (SP) of CC and the peak ratio (PR) of CC band height to C-C-C band height with a spatial resolution of 10 µm on both types of RCF were evaluated, averaged, and plotted as a function of dose. An achievable spatial resolution, clinically useful dose range, dosimetric sensitivity, dose uniformity, and postirradiation stability as well as the orientation, energy, and dose rate dependence, of both types of RCFs, were characterized by the technique developed in this study. Results: A spatial resolution on RCF achieved by SP and PR methods was~4.5 and~2.9 µm, respectively. Raman spectroscopy data showed dose nonuniformity of~11% in SP method and <3% in PR method. The SP method provided dose ranges of up to~10 and~20 Gy for EBT3 and EBT-XD films, respectively while the PR method up to~30 and~50 Gy. The PR method diminished the orientation effect. The percent difference between landscape and portrait orientations for the EBT3 and the EBT-XD films at 4 Gy had an acceptable level of 1.2% and 2.4%, respectively. With both SP and PR methods, the EBT3 and the EBT-XD films showed weak energy (within~10% and~3% for SP and PR methods, respectively) and dose rate dependence (within~5% and~3% for SP and PR methods, respectively) and had a stable response after 24-h postirradiation. Conclusions: A technique for micrometer-resolution dosimetry was successfully developed by detecting radiation-induced Raman shift on EBT3 and EBT-XD. Both types of RCFs were suitable for micrometer-resolution dosimetry using CRS. With CRS both lateral and depth resolutions on RCF were improved. The PR method provided superior characteristics in dose uniformity, dose ranges, orientation dependence, and laser effect for both types of RCFs. The overall dosimetric characteristics of the RCFs determined by this technique were similar to those known by optical density 5238 scanning. The CRS with the PR method is advantageous over other the traditional scanning systems as a spatial resolution of <10 µm on RCF can be achieved with less deviations.