Modelling insulin secretion as a function of peripheral C-peptide levels by mathematical deconvolution is widespread. However, the measurement resolution for successful deconvolution and high cost of C-peptide assays means measurement of insulin secretion can only be undertaken in small scale research endeavours. This research models the nature of insulin secretion (UN) during the pathogenesis of type 2 diabetes.A proportional-derivative UN model is based on the physiological, closed-loop insulin secretion response to increasing glucose (ϕD) and glucose excursions (ϕP). A total of 204 dynamic insulin sensitivity and secretion test (DISST) data sets from 68 participants in a 10-week dietary intervention trial were used to determine ϕD and ϕP values. The resulting gain values are used to classify subjects and thus the evaluation of UN over increasing insulin resistance.Participants with impaired fasting glucose (G0>5.56mmolL-1) had a lower median ϕD value that becomes almost equal to ϕP. In contrast, NGT participants (G0<5.56mmolL-1), ϕD that tended to be much greater than ϕP. Thus, as the metabolic state of a participant moves from NGT to pre-diabetes, the participant is loses first phase insulin burst secretion. The resulting gains are classified by easily measured basal glucose.The simplicity of this PD UN model in a DISST model framework provides clear relationship between the UN profile and the readily available metabolic state of each participant. These relationships could significantly improve the cost and resolution of model-based tests like the DISST.