Plasma levels of lipoprotein [a] (Lp[a]) are associated with increased risk of coronary artery disease and show an inverse correlation with apolipoprotein[a] (apo[a]) molecular weight. We determined Lp[a] levels and apo[a] phenotypes in 171 cases with preclinical extracranial carotid atherosclerosis as ascertained by B-mode ultrasound and in 274 control subjects free of carotid atherosclerosis. Lp[a] protein levels measured by enzyme-linked immunosorbent assay ranged from 4 to 361 /ig/mL in cases and from 2 to 392 jug/mL in controls, but median levels of Lp[a] were higher in cases than in controls (51 Aig/mL versus 33 /ig/mL,p<.003). In both groups, all 11 apo[a] polymorphs that are resolved by the procedure used were present, resulting in 43 and 39 different apo[a] phenotypes in cases and controls, respectively. An inverse relation between apo[a] polymorph size and Lp[a] level was observed in both cases (r=&#151; 0.49,p<.001) and controls (r=&#151;0.34, /><.001). Apo[a] phenotype distributions were similar in cases and controls. However, in 17 phenotypes with three or more subjects per group, the difference of mean Lp[a] concentrations between cases and controls was 32±36 /ig/mL (mean±SD). Thus, the higher Lp[a] levels in cases were not associated with a greater prevalence of small apo[a] polymorphs. Stepwise logistic regression analyses of known risk factors for coronary heart disease showed that plasma Lp[a] concentration was an independent predictor of case-control status, while Lp[a] phenotype was not, irrespective of the presence or absence of Lp[a] concentration in the model. These findings confirm the inverse correlation of apo[a] size with plasma Lp[a] levels but imply that sequences at the apo[a] gene locus other than those determining the number of apo[a] kringle type 2 units are involved in the increased expression of apo[a] in subjects with early atherosclerosis.