An optical dating study of some quartzose sediments from northern and southern Australia, north Africa and north-western Europe has revealed that palaeodose (P) underestimates of 10-40% are obtained when an extended-duration preheat of 160°C for 16 h is used. For these samples in the 0-60 ka age range, a preheat of 220°C for 5 min produces the correct P, as inferred from the concordance with thermoluminescence (TL) palaeodose determinations on the same sedimentary sample or on burnt flint. Independent support for use of the 220°C preheat is given by the agreement between optical ages younger than 30 ka and 14C age determinations on associated charcoal. The deleterious effect of the 160°C preheat is illustrated by growth curves of optically stimulated luminescence (OSL) vs added dose. These curves show that the 160°C preheat induces a dose-dependent increase in sample sensitivity, which may be related to activation of the pre-dose mechanism. The 160°C preheat also causes low-dose supralinearity, although the supralinearity correction is insufficient to reduce significantly the degree of P underestimation. A normalization method based on preheat-induced sensitivity changes of the 110°C TL peak, albeit limited by saturation effects, warrants further investigation. © 1994.