Epidemiologic and genetic data suggest an inverse relationship between plasma high density lipoprotein (HDL) cholesterol and the incidence of premature coronary artery disease. Some of the defects leading to low levels of HDL may be a consequence of mutations in the genes coding for HDL apolipoproteins A-I and A-II or for enzymes that modify these particles. A proband with plasma apoA-I and HDL cholesterol that are below 15% of normal levels and with marked bilateral arcus senilis was shown to be heterozygous for a 45-base pair deletion in exon four of the apoA-I gene. This most likely represents a de novo mutation since neither parent carries the mutant allele. The protein product of this allele is predicted to be missing 15 (Glu146-Arg160) of the 22 amino acids comprising the third amphipathic helical domain. The HDL of the proband and his family were studied. Using anti-A-I and anti-A-II immunosorbents we found three populations of HDL particles in the proband. One contained both apoA-I and A-II, Lp(A-I w A-II); one contained apoA-I but no A-II, Lp(A-I w/o A-II); and the third (an unusual one) contained apoA-II but no A-I. Only Lp(A-I w A-II) and (A-I w/o A-II) were present in the plasma of the proband's parents and brother. Analysis of the HDL particles of the proband by sodium dodecyl sulfate-polyacrylamide gel electrophoresis revealed two protein bands with a molecular mass differing by 6% in the vicinity of 28 kDa whereas the HDL particles of the family members exhibited only a single apoA-I band. The largely dominant effect of this mutant allele (designated apoA-ISeattle) on HDL levels suggests that HDL particles containing any number of mutant apoA-I polypeptides are catabolized rapidly.