The eating quality of porcine meat is supposed to increase with raising intramuscular fat content (IMF). The aim of this study was to investigate the use of acoustic parameters obtained by spectral analysis of ultrasound echo signals to non-destructively predict intramuscular fat content of porcine longissimus muscle. Ultrasound data acquisition was performed at slaughter with a clinical B-mode device at 115 warm carcasses. Acoustic parameters quantifying attenuation, backscattering and cepstral structure were calculated off-line from the recorded primary ultrasound echo signals. The IMF content could be predicted with mean standard error of prediction of 0.36 % IMF, and the ultrasound parameters were able to correctly classify nearly 80 % of the carcasses into HIGH and LOW level fat class, respectively. To generally extend the potential use of ultrasound spectral analysis for tissue characterisation and to improve the IMF estimation in pigs, conclusions are drawn with respect to the further technically develop the data acquisition system. Besides, relationships between sensory properties, meat quality indicators, and carcass features, respectively were investigated.