Multivariate analysis was applied to test the adequacy of pooling samples versus single units as a sampling strategy to estimate dietary intake and risk assessment of chemical elements in nonalcoholic beverages. The contents of 18 minerals and trace elements (Cr, Mn, Cu, Zn, As, Se, Mo, Sr, Co, Cd, Sn, Pb Fe, Mg, Ca, P, Na, and K) in fruit juices and nectars were determined by inductively coupled plasma–optical emission spectrometry and inductively coupled plasma–mass spectrometry. Arsenic speciation was done with high‐performance liquid chromatographer‐inductively coupled plasma–mass spectrometry. The data obtained in pooled samples and single units were then studied by analysis of variance and least significant difference tests, Spearman's correlation, principal component analysis, and cluster analysis. Analysis of variance and least significant difference tests were used to evaluate analytical data from pooled samples and single units. Values of individual units and pooled samples were statistically different with the exception of Se (P < .05), illustrating pooling as an inadequate strategy. Spearman's correlation displayed significant correlations between the following pairs: P‐K (0.924), Mo‐K (0.888), Mo‐P (0.876), and Ca‐Mg (0.846), indicating that chemical elements could be from the same source, having a natural occurrence, or the same exposure sources. By applying principal component analysis and cluster analysis, it was possible to classify juices and nectars by fruit type and geographical origin. It was observed that 2 principal components accounted for 59% of the total variance in the data. Cluster analysis classified samples into 5 clusters. Combined chemometric tools are suited to select appropriate laboratory sampling strategies for risk assessment. Application of chemometric methods to analytical data may be useful to group samples by similar characteristics for the purpose of Total Diet Studies.