Abstract Age determination of Atlantic redfish (Sebastes spp.) has proven difficult and has led to inconsistent age and growth estimates in the past. Using otoliths of the two major commercial species, golden redfish (Sebastes marinus) and deep-sea redfish (S. mentella), a series of exchange schemes was carried out to assess bias and precision of age readings between four readers and between two preparation methods. Considerable bias between readers and moderate precision were observed for the S. marinus readings, especially for ages >20 years, with coefficients of variation (CV) of 7.7–12.0% and average percent error (APE) of 5.4–8.5%. Agreement between readers increased from 17–28% to 45–61% when allowing deviations of ±1 year, and to 80–92% with ±3 years tolerance. The age of S. marinus determined from broken and burnt otoliths was estimated to be slightly lower than when the age of the same individuals was determined from thin-sectioned otoliths. The bias and precision estimates obtained from the S. mentella material were generally poorer than for S. marinus (CV 8.2–19.1%, APE 5.8–13.5%), but similar to reported values for other long-lived fish species. Better than 50% agreement was only achieved with ±3 years tolerance. Growth rates differed significantly between species, confirming slower growth for S. mentella. For S. marinus, only one reader comparison revealed significantly different growth functions, whereas almost all S. mentella reader pairs showed significant differences in growth curves. Section and break-and-burn readings of S. marinus did not differ significantly. Average ages of around 9–10 years were determined for juvenile S. mentella 24–30 cm long, which were likely to have migrated from East Greenland into the Irminger Sea, based on earlier observations. As some of the error in the age determinations presented could be attributed to interpretation differences between readers, further intercalibration of redfish ageing is urgently needed in order to provide consistent input data for stock assessment.