The paper uses utility theory to investigate how much should be spent to avert all costs from an industrial accident apart from direct human harm. These "environmental costs" will include those of evacuation, clean-up and business disruption. Assuming the organisation responsible will need to pay such costs, the difference between its expected utility with and without an environmental protection system constitutes a rational decision variable for whether or not the scheme should be installed. The value of utility is dependent on the coefficient of relative risk aversion, "risk-aversion" for short. A model of an organisation's decision-making process has been developed using the ABCD model, linking the organisation's assets, A, the cost of the protection scheme, B, the cost of consequences, C, and the expected utility difference with and without the scheme, D. Increasing the organisation's risk-aversion parameter will tend to make it less reluctant to invest in a protection system, but can bring about such investment only when the scheme is relatively close to financial break-even. For such borderline schemes, the amount the organisation is prepared to spend on the protection system will rise as the risk-aversion increases. The ratio of this sum to the break-even cost is named the "Limiting Risk Multiplier", the maximum value of which is governed by the maximum feasible value of risk-aversion. However, the mathematical model shows that increasing the risk-aversion will reduce the clarity of decision making generally. Although the reluctance to invest in a protection scheme may change sign and turn into a positive desire to invest as the risk-aversion increases, the absolute value of this parameter is a continuously decreasing function of risk-aversion, tending asymptotically to zero. As a result, discrimination will gradually diminish, being lost altogether at the "point of indiscriminate decision". Here the decision maker will be able to distinguish neither advantage in installing the scheme nor disadvantage in installing its inverse. There is a close correspondence between this mathematically predicted state and that of panic, where an individual has become so fearful that his actions become random. The point of indiscriminate decision provides a natural upper bound for the value of risk-aversion. This bounds the Limiting Risk Multiplier in turn, and so sets an objective upper limit on the amount that it is rational to spend on an environmental protection system. (C) 2010 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.