Microelectrode recordings (MER) involve insertion of an electrode, approximately 50 micrometers at the tip, into deep brain structures and recording the electrical activity. It is commonly used in surgeries to accurately determine the location of a target for deep brain stimulation (DBS). A common target for this type of procedure is the subthalamic nucleus (STN), identified by a unique spiking pattern and a change in the background noise level compared to surrounding structures. This change in background noise level indicates that the noise is composed of neuronal sources. This thesis aims to develop a model to determine what extent the volume of neurons around the electrode contribute to an MER. The potential usefulness of the model as a biomarker for the behaviour of the STN is then explored. The model for the STN MER involves simulating thousands of neurons by assuming they follow a renewal process. This assumption requires that the timing between a single neuron's spikes (inter-spike interval-ISI) are independent and identically distributed (IID). The model is tuned to intraoperative recordings to determine the best simulation parameters. To investigate the usefulness of the model as a biomarker of STN behaviour, the IID assumption is relaxed by introducing synchronization between neurons or changing their spike timings to be driven by a dynamical model of the basal ganglia. Fitting the parameters of the renewal process model to these extended models is then used to see if they can reliably describe the new behaviour. The results show firstly that a volume of ~1mm 3 , or on the order of 10,000 neurons, are required to simulate an STN MER that best describes patient data. This result indicates that the background noise of MERs is in fact partly caused by neuronal sources. The speed of the renewal model, faster than real time, allowed many simulations (Ο(10 6)) with different parameters to be created for tuning and verification of the model. The Weibull distribution is used as a parameterized ISI distribution for the renewal model and was found to reliably describe the simulations when the IID assumptions were relaxed. The first example of this was when neurons had synchronized firing times. In this case the Weibull distribution reliably fit the underlying ISI distribution for different realizations of the same simulation parameters. Finally, the thesis shows that using a neural mass model of the basal ganglia to generate the STN firing times, that the renewal model can differentiate between certain cortical inputs. This motivates future investigation into using the parameters from fitting a renewal model to an MER as biomarkers.