In most studies, the assessment of lesion "regression" is based on comparisons of group means between the "progression" and regression groups. This comparison depends on the assumption that the extent and distribution of lesions produced by the end of the lesion-induction period in the regression animals are equal to those observed in the progression group. To determine whether significant regression of lesions occurs in an individual regression animal, it is necessary to obtain a measure of the lesions produced in these animals at the end of the lesion-induction period. We achieved this goal by developing models using multiple stepwise regression analysis that related steady-state serum cholesterol and apolipoprotein B and A-I concentrations measured during a lesion-induction period in 27 rhesus monkeys fed an atherogenic high-saturated-fat/high-cholesterol diet for 2 years. The models were developed to estimate the percent of intimal surface with lesions, the esterified cholesterol content (micrograms/cm2) for the artery segments, and three histomorphometric measures (mean intimal thickness, mean maximal intimal thickness, and mean percent stenosis) for the coronary arteries. In these models, multiple R2 ranged from 0.42 to 0.74 for the aortas and peripheral arteries, indicating that approximately one half to three fourths of the variance in lesions was accounted for. For the three histomorphometric measures in coronary arteries, however, the multiple R2 was 0.27 or 0.28, indicating that only approximately one fourth of the variance in lesions was accounted for.(ABSTRACT TRUNCATED AT 250 WORDS)