The relationship between an ordinal (Likert) and a continuous (visual analog) measure of pain was investigated in 400 long-distance runners experiencing delayed-onset muscle soreness. Subjects completed a 100-mm visual analog scale (VAS) and a 7-point Likert scale twice a day for 5 days, starting at 9:00 PM on the evening of the run. The relationship between scales was modeled by calculating the median VAS for each Likert score and the modal Likert score for each point on the VAS. Standardized means were calculated by dividing the total of 5 day scores by the standard error of 5 day scores for all subjects. The relationship between scales was approximately linear, but VAS scores recorded concurrently with each Likert score varied enormously. VAS responses for a Likert score of zero were lower than expected, apparently because subjects chose zero only if they were completely free of soreness. The standardized mean of the Likert scale was higher (34.1 [SD 20] versus 30.4 [SD20] p <<< .0001), suggesting greater responsiveness. The Likert scale can be recommended as a method of measuring muscle soreness, but researchers should not anchor the lowest score to zero pain.