Using the Utah scatter phantom with media of different densities and activity proportions, we tested the accuracy of quantitation in the two-dimensional (2-D) and three-dimensional (3-D) modes of acquisition of the ECAT EXACT HR+ tomograph, The influence of radioactive sources outside the held of view (FOV) was also investigated, using different bed positions. We measured the activities in regions of interest drawn in the reconstructed images and compared the means through several planes versus the activities expected from well counter measurements. The 3-D scatter correction algorithm successfully restored the activity concentration in the various chambers of the phantom, with relative errors between -10.4 and +12.0%, slightly larger than those observed in 2-D (-8.6 to +5.6%) but similar to values when no activity was out of the FOV (between +4.0 and +4.6%). The activity out of the FOV increased the relative errors and introduced nonuniformities along the axial direction, The attenuation coefficient of the different materials also influenced the results: negative activity values were observed for high attenuation coefficient regions of the 3-D scatter corrected images.