People do not start with a blank slate when they hear risk-communication messages. All such messages are processed through existing knowledge structures and understanding. Hence, to design effective and reliable risk-communication materials one must understand the state of people's knowledge--correct and incorrect--about an issue. We developed a simple "mental model" of what people minimally need to know to make informed decisions about field-related issues. Then we performed studies to explore how and to what extent respondents of various groups understood physical properties of 60-Hz electric and magnetic fields. Actual knowledge of respondents was then compared with the predicates of the model. Electrical engineering juniors and semi-technical employees of utilities displayed a good command of most of the concepts in the simple model, but little awareness of the limits to their knowledge. Lay respondents correctly knew only a few of the simplest elements of the model, but they displayed a much greater awareness of the limits to their knowledge. Both lay and semi-technical respondents were found to share several misconceptions. On average, they correctly rank-ordered some common field-exposure conditions by field strength, but they could not differentiate between electric and magnetic fields and could not differentiate among field strengths associated with different appliances. Most respondents dramatically underestimated the range of actual field strengths. Many respondents understood that field strength decreases with distance from a source, but they underestimated the rate of decrease. In contrast to X-rays and microwaves, which respondents appeared to think about in rather similar terms, 60-Hz fields were not thought of as being highly similar to any other agent, although the closest parallels were found with ultrasound. Changes in mood, thought, and behavior, and the existence of an "electrical aura," were all seen as plausible results of exposure to a 60-Hz field. Although lay respondents displayed a variety of incomplete and confusing ideas, most of these ideas probably do not pose significant obstacles to the learning of a correct, simple, mental model.