The purpose of this thesis was to estimate genetic parameters for ultrasound-measured carcass traits in a multi-breed sheep population. Field data collected between 1997 and 1999 from 26 producers across Ontario were used in the analysis. Data were collected for three measurements: loin depth, backfat depth and loin width using the Ultrascan 50. Genetic and phenotypic parameters including heritability, genetic and phenotypic (co)variances and resulting correlations were estimated assuming an animal model. Heritabilities were estimated as 0.29, 0.29 and 0.26 (weight-constant) and 0.38, 0.35 and 0.30 (age-constant) for the traits loin depth, average backfat thickness and loin width respectively. Genetic improvement in carcass traits can be made through selection based on these ultrasound-measured traits. Data were also collected from an experiment with 38 Rideau-Arcott X Dorset lambs from the New Liskeard Agricultural Research Station to examine the accuracy of the Ultrascan 50 to measure tissue depth. Repeated ultrasound measurements were recorded for a total of 7 replicates per trait per animal. Within and between animal variances were calculated using ANOVA. Pearson correlation of 0.93 for loin depth was calculated between the ultrasound-measured trait and the same measurement on the carcass. Ultrasound-measured traits should be a valuable tool in improving meat quality in the sheep industry based on the results of this study.