We aimed to develop a simple, automated system to measure feeding motivation in cattle based on behavioural economics and using a natural (walking) operant response. A feed delivery system and feeder were constructed with two feed access bays. Presence at one bay (detected by infra red beams) activated feed delivery in the second bay. Access to the feed required the animal to walk around a fence panel extending out from the feeder. Continued access to feed required the animal to cycle back and forth between the bays, with the distance walked controlled by the length of the barrier. A software interface automatically recorded each feeding event and controlled the amount of feed delivered. The system was tested and used to examine the sensitivity of cattle to feed reward size (30 and 45 g) and walking distance up to 24 m. After a period of training and adjustment to a concentrate ration, feeding motivation of six yearling Angus steers was determined. Each animal was tested at each of five distances (costs) and the number of feed rewards over 23 h was measured. There was a 48 h recovery period between tests. Three animals received rewards of 30 g at all five distances (randomised) before being tested at the 45 g reward amount, with the other animals rewarded in reverse order. Demand function analyses of the data showed that Omax, the maximum distances walked, were 4.5 ± 0.5 and 13.2 ± 25.7 km for the 30 g and 45 g reward sizes, respectively. However, at 45 g the curve was relatively inelastic and the estimate of Omax was unreliable. An analysis of covariance revealed a significantly lower number of feed rewards (P<0.05) at 30 g compared to the 45 g indicating that cattle could detect the small difference in reward size.