The accuracy of the multiple pass 24 h recall for assessment of habitual energy intake in pre-school children is unclear. The primary aim of this study was to assess its accuracy by comparison with measurement of total energy expenditure by doubly-labelled water in a representative sample of forty-one 3–4-year-olds. The recall method was well tolerated by subjects and was administered quickly and easily. However, it produced estimates of energy intake which significantly exceeded measures of total energy expenditure from doubly-labelled water, mean paired difference 660 kJ/d (P<0·01). Agreement between the two methods was poor at the individual level: limits of agreement 660+/-3018 kJ/d. Error and imprecision in both methods contribute to individual differences, but the present study suggests that the multiple pass 24 h recall does not provide accurate estimates of dietary energy intake in individual children.