Two studies examined the effectiveness of a variety of commonly used likelihood scales (dichotomous scale, five-point verbally-labeled scale, odds scale, percentage scale, etc.) for assessing perceptions of personal susceptibility to health and safety risks. Two direct evaluation criteria (subjects' ratings of how easy it was to use a scale and of how well the scale reflected their feelings) and three additional measures of validity and reliability (the agreement of scale-derived ranks with a direct ranking, the magnitude of correlations between risk perceptions and appropriate risk factors, and the stability of scores over time) were used to compare the scales. Increasing the number of scale categories did not necessarily improve performance. In fact, a scale with seven verbally-labeled categories performed as well as or better than other scales on all evaluation criteria. These data may help health education researchers in selecting scales to measure perceptions of susceptibility to harm.