Risk perception may be influenced by a number of factors, such as unfamiliarity, lack of control, perceived consequences, and hazards being seen as catastrophic and having risk for future generations. Risk perception researchers have typically used such investigator-selected characteristics to assess hazards. In the first study reported here, the repertory grid method was used to elicit the terminology that subjects (n = 30) use to distinguish between 30 different chemicals. The data were submitted to generalized Procrustes analysis. The first principal axis of the resulting consensus plot separated the chemicals ranging from "poisonous or toxic," "harmful or dangerous," and "sounds negative" at one end, to "positive effect on health," "often present in food nowadays," and "sounds positive" at the other end. The second principal axis ranged from "familiar with or knowledge of" and "chemical" to "natural." A second study (n = 226) was carried out to look at the general validity of the results of the repertory grid interviews using a fixed questionnaire. The data were submitted to principal components analysis and internal preference mapping. The first principal component ranged from "safe" and "healthy" at one end, to "poisonous" and "harmful" at the other end. The chemicals also separated in terms of "familiar," "chemical," and "natural." All three methods of data collection and analysis yield essentially similar results.