The presence of large predators entails a range of costs and some benefits for the communities that live alongside them. The cost in terms of the value of livestock lost to predation is well known, but this represents only a part of the costs that people experience, as it does not account for non-market costs such as fear, avoidance behaviours and threat to human life. We quantify the total cost of lion presence for agro-pastoral communities in Zimbabwe using economic valuation techniques. The total perceived value of one additional lion was estimated to be negative US$180 per person per year, which is several hundred times larger than the market value of the average loss of livestock per household. If making simplifying economic assumptions, this discrepancy reveals the magnitude of the non-market costs associated with the presence of large predators. The disvalue of lion presence was linked to fear, ecocentric values, wealth and trust in compensation institutions. We also demonstrate the importance of considering heterogeneity in preferences within communities; segments of the population did not have net negative value attached to lions, and some disliked monetary compensation. We also estimate the willingness-to-accept for two human-wildlife conflict mitigation programs in terms of acceptable increases in lion numbers, as opposed to monetary units which is the conventional approach in economic valuation studies. We argue that estimating value in terms of biodiversity outcomes should be used more widely. We also demonstrate the distorting effect of distrust towards compensation in choice experiments and argue that economic valuation methods employing choice experiments should control for this distrust when estimating willingness-to-pay. These discoveries have relevance for a wide range of situations where potentially dangerous wildlife species co-exist with people.