Intuition suggests that large, dangerously venomous snakes should be relatively invulnerable to predators, but actual data on this topic are very scarce. We examined sources of mortality in two species of large elapid snakes (eastern brownsnakes, Pseudonaja textilis, and common blacksnakes, Pseudechis porphyriacus) in the Murrumbidgee Irrigation Area in southeastern Australia. Most attacks on snakes were by people and fetal cats. Over a three-year period, at least 19% of our radio-tagged brownsnakes were attacked, and eight snakes (14% of this population) were killed. We also examined road-killed snakes in the general area, and surveyed the public about where and when they encountered snakes, and how both they and the snakes responded to those situations. These questionnaire results suggest that about half of the snakes seen by people (whether in towns or on farms) were approached and about one-third were killed. Most of the mortality involved adult male snakes during the breeding season (spring). Snakes encountered by women were more likely to be killed than were those found by men. The response of snakes to encounters with people was quantified (by means of >500 encounters on regularly-walked transects) to compare with people's reports of snake responses. This comparison revealed that people are at least 20 times more likely to advance toward a snake, and 100 times more likely to attack, than is a snake to advance on, or attack, a person. People grossly over-rate the degree of aggression shown by these animals, and frequently misinterpret defensive displays as attacks. Furthermore, people are unaware of most of the snakes they encounter (especially those which use crypsis to avoid detection), and this biases the results towards over-estimation of aggressive responses. Attacks by humans on snakes are largely unjustified, and we need public education to reduce both personal risk and the slaughter of wildlife.