Discrepancies between PET/CT and PET/MRI scanners can affect the determination of amyloid beta (Aβ) deposition thresholds in patients with cognitive impairment. This study aimed to identify these differences and propose a calibration method to standardize Aβ quantification across imaging modalities.A total of 133 patients with cognitive impairment underwent Aβ PET imaging and were divided into four groups: a head-to-head PET/CT and PET/MRI cohort (group A, n = 6), an independent PET/CT cohort (group B, n = 48), an independent PET/MRI cohort (group C, n = 79), and another independent PET/MRI cohort (group D, n = 10). Standardized uptake value ratios (SUVR) of global cortical target (CTXsuvr) and centiloid (CL) values were compared within group A and between groups B and C. A whole cerebellum (WC)-referenced SUVR method was used to calibrate CL values in group C, with verification in group D.CTXsuvr values were significantly higher in PET/MRI than in PET/CT in both group A (P < 0.05) and group C versus group B (P < 0.001). Aβ-negative/positive cases showed mean ± variance of CTXsuvr as 1.023 ± 0.104/1.479 ± 0.203 in group B and 1.146 ± 0.100/1.743 ± 0.254 in group C, with cutoffs of 1.140 (CL = 20) and 1.401 (CL = 60), respectively. WC-referenced calibration adjusted PET/MRI cutoff to 1.132 (CL = 19) in group C, aligning it with PET/CT thresholds and validated in group D.WC-referenced SUVR calibration effectively mitigates differences in Aβ thresholds between PET/CT and PET/MRI, enhancing Aβ quantification standardization in multi-modal imaging.© 2025. The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.