One way to assess a new dating method's reliability is by comparing its results with those from well established, independent techniques. A controlled test of the electron spin resonance (ESR) dating method as it is currently being applied to teeth was attempted for the time range 100-250 ka, beyond that of 14C, at the archaeological site of La Chaise-de-Vouthon (Charente, France). Although absent in modern enamel, a single ESR signal with g = 2.0018 in fossil tooth enamel hydroxyapatite increases in amplitude with increasing irradiation doses. ESR ages are derived from the ratio of the AD, the radiation dose needed to produce the observed ESR signal, relative to the natural, environmental dose rate (ED) experienced by the tooth after deposition. Since the age depends on the uranium (U) uptake history assumed, three ages are calculated assuming: (1) early U uptake (EU); (2) continuous (linear) uptake (LU); (3) recent uptake (RU). Generally, the LU age agrees best with known ages determined by other methods, although the RU model is better for some teeth. ESR dating assumes that the fossil has not suffered recrystallization or significant diagenetic alteration. In the preliminary test, three teeth were dated. In Bourgeois-Delaunay, a bovid molar associated with Palaeolithic artefacts was collected from layers dated at 101 ± 12 to 114 ± 7 ka by 230Th 234U dating of the over- and underlying stalagmitic floors. From Suard, two Equus teeth were collected from beneath a stalagmitic floor dating 112 ± 12 ka. ESR dating teeth significantly underestimated the true age for the teeth: the mean ESR ages range from 37 to 94 ka with standard errors of 2-6 ka, and good replicability. Although more teeth at La Chaise need to be tested to ascertain that the underestimation does not result from random variation commonly seen among teeth within one unit, the consistent underestimation suggests a fault in one of the assumptions underlying the dating method. The most obvious source of error lies in the difficulty in modelling the external γ dose. Only U leaching, not incorrectly modelled U uptake, would cause the underestimation. Diagenetic alteration may also cause anomalous fading, thermal instability, variation in k, or ESR signal suppression. More study into the effects of diagenesis alteration on enamel ESR signals is needed, as is a reevaluation of the mean signal life and α efficiency for several more enamel samples.