Determination of total content (сΣ) of similar analytes recalculated to the standard substance Xst is a commonly used but metrologically incorrect measuring procedure leading to significant systematic errors (B-type uncertainties). Interval estimate сΣ is an alternative method (V.I. Vershinin et al., 2016) not requiring recalculation to Xst and, in difference from calculating total indices, not depending on the nature and ratio of the sought group constituents present in the sample. Such estimates are used for evaluating total content of antioxidants (AO) in foodstuffs. However, the dependence of these estimates on the choice of group reagent and signal measuring procedure is uncertain. To address this question, model mixtures of AO with known сΣ values ranging from 10–5 to 10–4 mol/L were prepared and examined. Generalized signals were measured spectrophotometrically using Folin-Ciocalteu (F-C) and FRAP methods, and traditional and interval estimates of сΣ were calculated and compared. The true values of сΣ were within the respective intervals for all model mixtures; in case of F-C method the intervals were wider and shifted towards higher сΣ values. The FRAP procedure was modified by using AO-concentrations in mol-eq/L, reducing the reaction time and substituting the auxiliary reagent. This modification successfully leveled the sensibility coefficients of individual AOs and made the relative width of intervals three times as narrower. Modified FRAP method was used for group analysis of wines, black tea infusions and juices. Corresponding interval estimates of сΣ were obtained and compared. Unresolved issues and future research directions related to interval estimates are discussed.