Using personnel records from two firms in the banking industry, job duration models are estimated to examine separations in the context of banks based in Great Britain and Greece. We find that it is sustained, rather than instantaneous, performance that is linked to separations. In common with some earlier studies, we find qualified support for a u-shaped relationship between performance and separations – suggesting that poor matches are short-lived and that high-performance workers move on to other employment – but only in the case of the British data. Both of the banks under investigation experienced substantial reorganisation activity over the time period considered, and we find that the year following this was characterised by increased separation propensities. While most of our findings are consistent across the firms in the two countries studied, we find that single men are more likely than their female counterparts to quit in the bank based in Britain, but less likely to quit in that based in Greece. We offer some suggestions about why this should be the case.