Nonarticular tenderness was measured in 152 patients, 51 with rheumatoid arthritis, 50 with psoriatic arthritis, and 51 with human immunodeficiency virus infection. Three methods for assessing tenderness were used: a 14-site point count, scored tenderness at 10 sites (6 fibrositic and 4 control), and dolorimetry at the same 10 sites. The data from the 3 separate measures were converted into a common scale of standard deviation units for further analysis. There were strong correlations among the 3 measures. In particular, the scored tenderness at just 6 fibrositic sites provided as much information about the presence and severity of widespread tenderness as the other 2, more complex measures. At the interface between nontender and tender, the 2 methods based on palpation were significantly more sensitive to differences among individuals, than was dolorimetry. However, the palpation scales used did not evaluate different degrees of nontenderness. In 102 of the 152 subjects, there were 842 sites scored zero by palpation, and which showed widely different thresholds of tenderness by dolorimetry, significantly associated with diagnosis and sex. For screening and epidemiological purposes, scored tenderness at a limited number of accessible sites may be adequate and feasible, using the 18 point count of the new standard criteria as a gold standard for confirmation. For the assessment of generally acting factors affecting tenderness, dolorimetry is currently superior.