Fructose is credited with some advantages over sucrose: it causes less of an increment in plasma glucose and insulin response, and the taste is sweeter. We reevaluated the latter property with a new methodology (the "up and down" method adapted from Dixon) in 33 healthy subjects, 17 insulin-dependent diabetes mellitus (IDDM) patients, and 12 non-insulin-dependent diabetes mellitus (NIDDM) patients. Sweetening potency was determined over 2-3 test sessions in each subject. Results are expressed in percent as the relative sweetness (R) of fructose (F) over sucrose (S), taken as reference. In the first set of experiments, with a 30-g/L sucrose-water solution at pH 7, we found that R values were similar for healthy subjects (102 +/- 8%) and diabetic subjects (106 +/- 7%) (P less than .05). No significant difference between IDDM and NIDDM patients was observed. In a second set of experiments, performed in healthy subjects only, R was increased in acid water (114%; P less than .01), in lemon juice (136%; P less than .001), in water at 2 degrees C (130%; P less than .001), and in coffee at 2 degrees C (120%; P less than .02); mean values were decreased in grapefruit juice (77%; P less than .001), in water at 43 degrees C (88%; P less than .01), and in coffee at 53 degrees C (87%; P less than .001). We found that the test methodology had a very satisfactory intrasubject reproducibility (coefficient of variation [C.V.] less than 8%) but a very wide intersubject variability (C.V. congruent to 32%).(ABSTRACT TRUNCATED AT 250 WORDS)