Improved understanding of the genetic architecture of the proteome through studies of larger sample size, ethnic diversity, and advanced methods can facilitate the identification of causal mechanisms for complex traits. We conducted a comprehensive analysis of the common variant cis-regulatory genetic architecture of 4,665 plasma proteins or protein complexes ascertained using an aptamer-based technology from 7,213 European Americans (EA) and 1,871 African Americans (AA) from the Atherosclerosis Risk in Communities (ARIC) cohort study. We identified and fine-mapped 1,992 plasma proteins or protein complexes in EA and 1,605 in AA, with majority overlapping in EA, which had at least one significant single-nucleotide polymorphism (SNP) in cis region. Estimates of cis-heritability (cis-h2) for plasma proteins were similar across the two ethnic groups (median cis-h2 = 0.09 for EA and 0.10 for AA). Elastic-net based models for cis-SNP-based protein prediction produced high accuracy for the EA population (median R2/cis-h2=0.79), and notably for the AA (median R2/cis-h2 = 0.69), despite the much smaller sample size in the latter population. We illustrate the application of these models to conduct proteome-wide association studies (PWAS) for two related complex traits, serum urate and gout, and further conduct conditional analyses to interpret findings in the context of those from transcriptome-wide association studies (TWAS).