Luminescence dating is used extensively to provide absolute chronologies for Late Pleistocene sediments. Nowadays, most optical dates are based on quartz optically stimulated luminescence (OSL). However, the application of this signal is usually limited to the last ~100 ka because of saturation of the quartz luminescence signal with dose. In contrast, the feldspar infrared stimulated luminescence (IRSL) dose–response curve grows to much higher doses; this has the potential to extend the datable age range by a factor of 4–5 compared with quartz OSL. However, it has been known for several decades that this IRSL signal is unstable, and this instability often gives rise to significant age underestimation. Here we test against independent age control the recently developed feldspar post-IR IRSL approach to the dating of sediments, which appears to avoid signal instability. A physical model explaining our observations is discussed, and the method is shown to be accurate back to 600 ka. The post-IR IRSL signal is reduced by exposure to daylight more slowly than that from quartz and low-temperature IRSL, preventing its general application to young (e.g. Holocene) sediments. Nevertheless, this new approach is widely applicable (feldspar of appropriate luminescence behaviour is even more ubiquitous than quartz). These characteristics make this a method of great importance for the dating of Middle and Late Pleistocene deposits.