To assess the relative validity and reproducibility of the quantitative FFQ used in the Tzu Chi Health Study (TCHS).

The reproducibility was evaluated by comparing the baseline FFQ with the 2-year follow-up FFQ. The validity was evaluated by comparing the baseline FFQ with 3 d dietary records and biomarkers (serum folate and vitamin B12). Median comparison, cross-classification and Spearman correlation with and without energy adjustment and deattenuation for day-to-day variation were assessed.

TCHS is a prospective cohort containing a high proportion of true vegetarians and part-time vegetarians (regularly consuming a vegetarian diet without completely avoiding meat).

Subsets of 103, seventy-eight and 1528 TCHS participants were included in the reproducibility, dietary record-validity and biomarker-validity studies, respectively.

Correlations assessing the reproducibility for repeat administrations of the FFQ were in the range of 0·46–0·65 for macronutrients and 0·35–0·67 for micronutrients; the average same quartile agreement was 40%. The correlation between FFQ and biomarkers was 0·41 for both vitamin B12 and folate. Moderate to good correlations between the baseline FFQ and dietary records were found for energy, protein, carbohydrate, saturated and monounsaturated fat, fibre, vitamin C, vitamin A, K, Ca, Mg, P, Fe and Zn (average crude correlation: 0·47 (range: 0·37–0·66); average energy-adjusted correlation: 0·43 (range: 0·38–0·55); average energy-adjusted deattenuated correlation: 0·50 (range: 0·44–0·66)) with same quartile agreement rate of 39% (range: 35–45%), while misclassification to the extreme quartile was rare (average: 4% (range: 0–6%)).

The FFQ is a reliable and valid tool to rank relative intake of major nutrients for TCHS participants.