Aim: We developed a new method to design objective, risk-based surveillance systems for non-indigenous species of invertebrates, vertebrates and plants, which might be introduced to a natural area through an industrial project; here, we provide the invertebrate case study. The method addresses issues common to complex surveillance design problems: a statistical standard (e.g. power); information gaps; multiple targets of unclear identity; a large surveillance area of heterogeneous risk of invasion; integrating multiple sources of surveillance data; optimizing for cost. Location: Barrow Island, Western Australia. Methods: We mapped the surveillance area for risk to target surveillance activities. An expert group identified a set of exemplar species and identified and characterized a set of detection methods for each, such that all potential invaders would be detected. We devised multi-element surveillance systems to detect each exemplar to the design power (0.8), then integrated them to a single system that was optimized for cost. Results: The surveillance system was deployed on the island to specification over 1 year, then reviewed for redesign in a second period. Main conclusions: The new method provided practical, risk-based surveillance system designs that met application requirements and overcame complex issues common to many surveillance applications. A review of experiences from surveillance in the first year led to practical improvements and design efficiencies.