The aim of this study was to correlate three commercially available copeptin assays and their diagnostic accuracy in the differential diagnosis of the polyuria-polydipsia syndrome. Analyzed data include repeated copeptin measures of 8 healthy volunteers and 40 patients with polyuria-polydipsia syndrome undergoing osmotic stimulation and of 40 patients hospitalized with pneumonia. Copeptin was measured using the automated Brahms KRYPTOR, the manual Brahms LIA and the manual Cloud Clone ELISA assay. Primary outcome was the interrater correlation coefficient (ICC) and diagnostic accuracy in the polyuria-polydipsia syndrome of the three assays. In healthy volunteers, there was a moderate correlation for the KRYPTOR and LIA (ICC 0.74; 95% CI 0.07 to 0.91), and a poor correlation for the KRYPTOR and ELISA (ICC 0.07; 95% CI - 0.06 to 0.29), as for the LIA and ELISA (ICC 0.04; 95% CI - 0.04 to 0.17). The KRYPTOR had the highest diagnostic accuracy (98% (95% CI 83 to100)), comparable to the LIA (88% (95% CI 74 to 100)), while the ELISA had a poor diagnostic accuracy (55% (95% CI 34 to 68)) in the differential diagnosis of the polyuria-polydipsia syndrome. The KRYPTOR and LIA yield comparable copeptin concentrations and high diagnostic accuracy, while the ELISA correlates poorly with the other two assays and shows a poor diagnostic accuracy for polyuria-polydipsia patients. The current copeptin cut-off is valid for the KRYPTOR and LIA assay. Our results indicate that interpretation with other assays should be performed with caution and separate validation studies are required before their use in differentiating patients with polyuria-polydipsia syndrome.Trial registration: NCT02647736 January 6, 2016/NCT01940614 September 12, 2013/NCT00973154 September 9, 2009.