Riboflavin status is usually measured as the in vitro stimulation with flavin adenine dinucleotide of the erythrocyte enzyme glutathione reductase, and expressed as an erythrocyte glutathione reductase activation coefficient (EGRAC). This method is used for the National Diet and Nutrition Surveys (NDNS) of the UK. In the period between the 1990 and 2003 surveys of UK adults, the estimated prevalence of riboflavin deficiency, expressed as an EGRAC value > or = 1.30, increased from 2 to 46 % in males and from 1 to 34 % in females. We hypothesised that subtle but important differences in the detail of the methodology between the two NDNS accounted for this difference. We carried out an evaluation of the performance of the methods used in the two NDNS and compared against an 'in-house' method, using blood samples collected from a riboflavin intervention study. Results indicated that the method used for the 1990 NDNS gave a significantly lower mean EGRAC value than both the 2003 NDNS method and the 'in-house' method (P < 0.0001). The key differences between the methods relate to the concentration of FAD used in the assay and the duration of the period of incubation of FAD with enzyme. The details of the EGRAC method should be standardised for use in different laboratories and over time. Additionally, it is proposed that consideration be given to re-evaluating the basis of the EGRAC threshold for riboflavin deficiency.