Adaptive value-guided decision-making often requires weighing-up of costs and benefits of pursuing an available opportunity. Several brain areas, particularly in frontal- striatal circuits, have been reported to be important for this behaviour. However, little is currently known regarding how the decision variables are represented and evolve on the single cell and population levels across these circuits within the confines of a single behavioural paradigm. We developed a novel rodent cost-benefit "accept-reject" task, in which food-restricted rats choose whether or not to run to the end of a corridor to collect sucrose pellets based on the prospective reward (4 different levels, varied trial- by-trial, signalled by auditory cues) and effort cost (3 different levels, varied over blocks of trials, implemented with barriers placed in the corridor that needed to be scaled to reach the reward magazine). Behavioural data (n=12 rats, average 1119 trials per animal) demonstrate a positive effect of reward and a negative influence of effort on the likelihood of accepting an offer, without an interaction between the two. Outcome devaluation strongly reduces acceptance of any offer. A subset of rats (n=5) were also implanted with a bespoke driveable micro-electrode array targeting, anterior cingulate cortex (ACC), medial orbitofrontal cortex (mOFC), dorsomedial striatum (DMS), ventral pallidum (VP), and subthalamic area (STA) and electrophysiological recordings were collected as they performed the task. Analysis of spiking data (n=4 rats; average 119 cells per region per animal) suggests that individual neurons in the 3 targeted basal ganglia areas (DMS, VP and STA) encode reward and/or, to a lesser extent, effort with high fidelity during cue presentation, while neurons in the two cortical areas (ACC and mOFC) display weaker tuning to reward/effort across multiple task points. By contrast, decision is progressively signalled by mOFC, DMS and VP units after action initiation. Decoding analyses using all the units from individual areas in a given session (ensembles) revealed that reward is signalled with similar spatial and temporal characteristics to single units. By contrast, effort is represented by DMS ensembles in a time-distributed manner and is dependent on diverse underlying neuronal activity profiles. These results are consistent with a parallel and distributed system for processing cost-benefit decision variables in the frontal cortical-basal ganglia network.