In vivo studies on the glucose system often require its perturbation by an exogenous input of glucose, whereas glucose turnover is assessed by infusing a glucose tracer. The constant infusion represents the usual format of tracer administration, but it has no clear advantage other than simplicity. Here we propose a different tracer infusion format. It consists of infusing the tracer in parallel with unlabeled glucose so as to maintain a constant specific activity in the infusate. This protocol does not increase experimental complexity and provides new information on the glucose system in non-steady state by allowing reconstruction of the endogenous component of glucose concentration. This reconstruction only requires very general assumptions, such as tracer-tracee indistinguishability and mass conservation; in particular it is independent of the glucose model structure, i.e., number of compartments and their interconnections. A proof of the result is given for a general nonlinear model of the glucose system. The constant specific activity input is also advantageous for non-steady-state calculations, because it reduces the variation in the measured plasma glucose specific activity. The glucose system has served as the prototype, but the protocol is applicable to other blood-borne substances. The radioactive tracer case has been considered, but the same results apply to stable isotope tracers as well; in this case they also become relevant in a somewhat different context, i.e., kinetic studies in steady state.