Using the history of dingo-proof fences in Australia, I show that several key lessons were learnt by pastoralists, but later forgotten and had to be re-learnt. Each has application to current and future proposals to build predator-proof fences for conservation. 1. Feral predators kill and eat native fauna. 2. Predators go under or through fences, or climb fences. 3. Predator-proof fences must be maintained. 4. Maintenance is expensive and must be continued indefinitely. 5. Scalp bonuses do not control predators and are a waste of money. Despite some assertions from conservationists, there is no evidence that the motive for building a fence (conservation versus growing sheep) is any guarantee of better performance. Proponents of predator-proof fences (and funding agencies) need to ask two key questions. 1. What is the real objective of the fence, and how do we measure its success? 2. What happens if/when the fences are successful? Unless predators are controlled outside the fence, then the enclosure remains a captive breeding zoo which achieves little for conservation at a landscape-scale.