In a new approach to noninvasively diagnose coronary artery disease (CAD), auditory correlates attributed to blood flow turbulence have been associated with stenosed coronary arteries. These auditory components have been detected in diseased subjects by spectral estimation of diastolic heart sound recordings made on the surface of the chest. In this study, we investigated the dynamics of the process that produces these sounds in diseased subjects by applying the techniques of dimensional analysis. Our results indicate a difference in the "correlation dimension" between heart sounds of diseased and normal subjects. In particular, diseased subjects show a fractal dimension, implying the presence of a strange attractor and the possible existence of low-dimensional chaos in sounds associated with coronary artery disease.