Introduction The evaluation of children food consumption is a challenge due to errors and bias of measurement, as well as due to the lack of efficient dietary assessment methods to estimate the intake among this age group. The objective of this study was to assess the reliability and relative validity of a food-frequency questionnaire (FFQ) designed to estimate food intake of children from metropolitan area in Rio de Janeiro/Brazil.

Methods This study enrolled 107 children, both genders, aged 2 to 5-years-old. The FFQ's food list included 73 food types and eight portion size options. Four 24 h dietary recalls were used as the reference method. Total energy intake and 14 food groups were estimated based on four 24 h dietary recalls and two FFQ. The reliability was estimated by Spearman and intra-class coefficients (ICC). Analyses included quartiles using weighted κ coefficient. The relative validation was assessment by Spearman coefficients and Bland-Altman method among estimated food groups.

Results The means of first FFQ were significant higher than 2^nd FFQ for energy, cereals, sugar drinks, cookies, sweets, sausages, and meat. The Spearman coefficients ranged from 0.24 (bean) to 0.77 (infants foods). ICC ranged from 0.22 (bean) to 0.98 (lettuces). The weighted κ ranged from 0.20 (cereals) to 0.47 (fats). To relative validity, the Spearman coefficients ranged from 0.03 (meat) to 0.51 (infants foods).

Conclusion The FFQ evaluated may be a suitable instrument for evaluating children dietary intake based on food groups. However, the 24 h dietary recall provided the best estimation for young children.