Background. In pharmacology the clinical efficacy of a compound is generally strongly dependent on baseline characteristics of the test subjects. In recent HMG -CoA reductase inhibitor (statin) studies this was not so. However, these studies used subgroup analysis, and subgroup analysis may not be sensitive to demonstrate a linear relationship between baseline LDL-cholesterol levels and responsiveness to statin treatment. Methods. We used regression analysis of previously published statin data to assess relationship between baseline LDL-cholesterol and responsiveness to statin treatment. Also we looked into the data of recent statin studies for mechanisms that tend to make comparisons more similar. Results. Regression unlike subgroup analysis is able to demonstrate a highly significant (p<0.0001) positive correlation between baseline LDL-cholesterol and responsiveness to statin treatment. In recent randomized statin studies reporting independence between responsiveness and baseline levels, the following mechanisms tended to make comparisons more similar: (1) the use of unadjusted results, (2) no test for the left tail of the chi-square distribution, and (3) the use of unrandomized endpoint variables. Conclusions. Although patients with low LDL-cholesterol levels may tend to benefit from statins, the amount of benefit must be considerably less than that of patients with high baseline values. Better benefit-risk analyses are required for the patients with low LDL-cholesterol levels before consensuses should dictate their routine use.