The separation of six soy isoflavones (Glycitein, Daidzein, Genistein, Daidzin, Glycitin and Genistin) was approached by a 3(2) factorial design studying MEKC electrolyte components at the following levels: methanol (MeOH; 0-10%) and sodium dodecylsulfate (SDS; 20-70 mmol L(-1)); sodium tetraborate buffer (STB) concentration was kept constant at 10 mmol L(-1). Nine experiments were performed and the apparent mobility of each isoflavone was computed as a function of the electrolyte composition. A novel response function (RF) was formulated based on the production of the mobility differences, mobility of the first and last eluting peaks and the electrolyte conductance. The inspection of the response surface indicated an optimum electrolyte composition as 10 mmol L(-1) STB (pH 9.3) containing 40 mmol L(-1) SDS and 1% MeOH promoting baseline separation of all isoflavones in less than 7.5 min. The proposed method was applied to the determination of total isoflavones in soy germ capsules from four different pharmaceutical laboratories. A 2h extraction procedure with 80% (v/v) MeOH under vortexing at room temperature was employed. Peak assignment of unknown isoflavones in certain samples was assisted by hydrolysis procedures, migration behavior and UV spectra comparison. Three malonyl isoflavone derivatives were tentatively assigned. A few figures of merit for the proposed method include: repeatability (n=6) better than 0.30% CV (migration time) and 1.7% CV (peak area); intermediate precision (n=18) better than 6.2% CV (concentration); recoveries at two concentration levels, 20 and 50 microg mL(-1), varied from 99.1 to 103.6%. Furthermore, the proposed method exhibited linearity in the concentration range of 1.6-50 microg mL(-1) (r(2)>0.9999) with LOQ varying from 0.67 to 1.2 microg mL(-1). The capsules purity varied from 93.3 to 97.6%.