The thesis presents an original investigation into the feasibility and operational efficiency of a novel prototype robotic rotary (RR) incorporated into a low-input, pasture-based Australian dairy farming system. A world’s first high throughput automatic milking system was installed, co-developed and tested at the Elizabeth Macarthur Agricultural Institute site (Camden, NSW, Australia). Being the first farm internationally to tackle voluntary distributed cow traffic (cf. batch milking) the challenges and learnings were specific to the system utilisation and the stage of technological development of the Camden installation. The thesis focuses on the challenges associated with application of the technology, of which learnings will have an immense level of importance for the first commercial installations on farm and further development of the system. These challenges were; (1) investigating a coping mechanism (with and without extra feed) to ensure and maintain high quality milk collection and storage for periods of underutilisation of the systems’ capacity (with voluntary cow traffic) in the absence of an automatic plant cleaning function, (2) understanding the impact of premilking teat preparation on the incidence of unsuccessful milkings, to ensure that farmers make an informed decision prior to commencement of the RR (as purchase of the teat preparation module will be optional), and (3) potential implications of management strategies for incompletely milked cows on dairy layouts. During periods of underutilisation the operator can deactivate a proportion of bails to better match the demand and availability of milk harvesting bails. Thus, investigations were conducted to understand the impact of bail activation sequence, availability of feed and cow queue size on voluntary cow traffic and robotic throughput efficiency. It was found that overall the availability of a feed reward as cows entered the RR had a larger effect on cow traffic than bail activation sequence, although the number of cows present (voluntarily) at the yard also played a role. Furthermore, having a greater number of consecutive bails activated resulted in more robot operations being conducted simultaneously resulting in an increased harvesting efficiency. Premilking teat preparation is also known to impact on milk harvesting efficiency, and as this component of the technology will be optional an investigation was conducted to assess the effects of not using a premilking teat preparation device on attachment accuracy and milk removal characteristics. The teat cup attachment was more successful and faster when cows were subjected to the teat cleaning treatment. Cows milked after being exposed to teat cleaning treatment, with a short milking interval (< 8 h), had a higher peak milk flow, however no difference was observed in the average milk flow rate of individual cows. Whilst there was an impact on attachment success by the use of the premilking treatment, the overall level of success was still lower than desirable. With this in mind a study of different management practices of incompletely milked cows was conducted. The system showed no difference in attachment success between milking incomplete cows after a one- or three-hour interval. This suggested that there is a level of flexibility available in designing the dairy layout and that no significant advantage or disadvantage (with regard to subsequent success level) exists in drafting incomplete cows directly back to the pre-milking yard compared to offering them an opportunity to spend time on a feedpad prior to the second attempt. The results presented in this thesis will be invaluable in furthering industry understanding of management practices with the new milk harvesting technology, the RR. The contribution of these scientific investigations will be extremely important to the success of the development of the system, which is progressing closer to commercialisation.