The paper proposed the molecular recognition of pyruvic acid and L-lactate in whole blood and urine samples using three stochastic microsensors based on graphenes' pastes (graphene materials decorated with nanoparticles of TiO2Ag or TiO2Au, and reduced graphene oxide) modified with alpha-cydodextrin. The proposed sensors covered large ranges of concentration: 10(-15) to 10(-6) for the assay of pyruvic acid at pH = 3 (urine samples), and pH = 7.4 (whole blood samples), and 10(-14) to 10(-6) for the assay of L-lactate at pH = 3 (urine samples), and pH = 7.4 (whole blood samples). The limits of determination are of fmol/L magnitude order, the lower being recorded using the microsensors based on Ag-TiO2-GR (0.1fmol L-1 at pH 7.4, and of 10 fmol L-1 magnitude order for pH 3.0). The sensitivities of all microsensors are very high. The validation of the method was performed using urine and whole blood samples from patients diagnosed with diabetes. The results obtained showed that the proposed microsensors and method can be reliable used for screening tests of the whole blood and urine samples, contributing to fast diagnostic the diabetic patients. (C) 2018 The Electrochemical Society.