Background: Radiation fields encountered in proton therapy (PT) and ionbeam therapy (IBT) are characterized by a variable linear energy transfer (LET), which lead to a variation of relative biological effectiveness and also affect the response of certain dosimeters. Therefore, reliable tools to measure LET are advantageous to predict and correct LET effects. Fluorescent nuclear track detectors (FNTDs) are suitable to measure LET spectra within the range of interest for PT and IBT, but so far the accuracy and precision have been challenged by sensitivity variations between individual crystals. Purpose: To develop a novel methodology to correct changes in the fluorescent intensity due to sensitivity variations among FNTDs. This methodology is based on exposing FNTDs to alpha particles in order to derive a detector-specific correction factor. This will allow us to improve the accuracy and precision of LET spectra measurements with FNTDs. Methods: FNTDs were exposed to alpha particles. Afterward, the detectors were irradiated to monoenergetic protons, 4 He-, 12 C-, and 16 O-ions. At each step, the detectors were imaged with a confocal laser scanning microscope. The tracks were reconstructed and analyzed using in-house developed tools. Alpha-particle tracks were used to derive a detector-specific sensitivity correction factor (k s,i ). Proton, 4 He-, 12 C-, and 16 O-ion tracks were used to establish a traceable calibration curve that relates the fluorescence intensity with the LET in water (LET H 2 O ). FNTDs from a second batch were exposed and analyzed following the same procedures, to test if k s,i can be used to extend the applicability of the calibration curve to detectors from different batches. Finally, a set of blind tests was performed to assess the accuracy of the proposed methodology without user bias. Throughout all stages, the main sources of uncertainty were evaluated.Results: Based on a sample of 100 FNTDs, our findings show a high sensitivity heterogeneity between FNTDs, with k s,i having values between 0.57 and 2.55. The fitting quality of the calibration curve, characterized by the mean absolute percentage residuals and correlation coefficient, was improved when k s,i was considered. Results for detectors from the second batch show that, if the fluorescence signal is corrected by k s,i , the differences in the predicted LET H 2 O with respect to the reference set are reduced from 55%, 141%, 41%, and 186% to 4.2%, 6.5%, 5.0%, and 11.0%, for protons, 4 He-, 12 C-, and 16 O-ions, respectively.