A risk-based sensor placement methodology is proposed to solve the problem of optimal location of sensors to protect population against the exposure to, and effects of, known and/or postulated chemical, biological, and/or radiological threats. Risk is calculated as a quantitative value representing population at risk from exposure at standard exposure levels. Historical meteorological data are used to characterize weather conditions as the frequency of wind speed and direction pairs. The meteorological data drive atmospheric transport and dispersion modeling of the threats, the results of which are used to calculate risk values. Sensor locations are determined via an iterative dynamic programming algorithm whereby threats detected by sensors placed in prior iterations are removed from consideration in subsequent iterations. In addition to the risk-based placement algorithm, the proposed methodology provides a quantification of the marginal utility of each additional sensor. This is the fraction of the total risk accounted for by placement of the sensor. Thus, the criteria for halting the iterative process can be the number of sensors available, a threshold marginal utility value, and/or a minimum cumulative utility achieved with all sensors.