The validity of estimates of resource utilisation based on patient recall has not been firmly established. A comprehensive literature search was conducted to identify studies that used patient self-reports to derive estimates of healthcare resource utilisation, direct nonmedical costs and indirect costs. Previous work in this area was examined to determine the issues that were most important in determining whether patient self-reports lead to valid and unbiased estimates in cost-effectiveness studies. This study reviews and highlights areas where patient self-reports lead to reliable estimates and where their use may lead to erroneous conclusions. In particular, it is noted that patient recall of resource utilisation declines over time, that the salience of a treatment episode affects recall and that the perceived social acceptability of a condition and other confounding factors may influence patient reporting. Moreover, it appears that not all elements of healthcare consumption are recalled to the same degree: medication use tends to be recalled with less accuracy than hospitalisation. In terms of the potential impact on cost-effectiveness results, the main concern is with problems of validity rather than bias, although bias may occur when the results are applied to league tables.