The influence of signal averaging (n), repetition time (TR), and echo delay (TE) on systematic noise (cardiac, vascular, respiratory, and peristaltic ghost artifacts) and statistical noise (thermal effects) was determined in eight healthy volunteers and 57 patients. Systematic noise was the dominant factor degrading abdominal magnetic resonance (MR) images. Signal averaging was the primary determinant of both statistical and systematic image noise, fitting a power function (n)b with b = 0.44 and -0.42, respectively, close to the expected b = -0.5 power function. All types of ghosting showed the same sensitivity to signal averaging. Normalized systematic noise increased slightly with TR (b = 0.16) and increased markedly with TE (b = 0.40). These data indicate that the short TR, short TE technique is a powerful method for reducing motion artifacts on breathhold images and can be combined with signal averaging to further suppress artifacts, improve signal-to-noise ratio, and maximize anatomic resolution.