This study discusses the values of equivalent dose (De) determined for samples taken for optical dating from a loess section at Loveland, Iowa, North America. A modified single-aliquot regenerative-dose (SAR) procedure was applied to quartz grains of 35–View the MathML source diameter which were isolated using hydrofluorosilicic acid (H2SiF6). ‘Preheat plots’ were constructed to select appropriate De values for use in subsequent age determinations. The De values of the younger (Peoria loess) samples varied little with preheat temperatures between 160 and View the MathML source. However, the older (Pisgah Formation) samples showed a decrease in De values with increasing preheat temperature, by as much as a factor of two. Interestingly, both younger and older samples could recover a known laboratory administered radiation dose for a range of preheat temperatures. These apparently contradictory findings present a dilemma regarding which De values are the most appropriate to use for dating the samples. The behaviour of the samples is examined to investigate the cause of the variations in De values with preheat temperature for older samples. In addition, two modified SAR procedures were tested in an attempt to generate more reproducible De values. The data from the three SAR methods are examined, and the question of how to discriminate between De values is discussed. This study demonstrates that preheat temperature can play a significant role in the determination of De values. The use of plots of normalised luminescence signal (Lx/Tx) versus preheat temperature is proposed to supplement dose recovery tests and preheat plateau tests.