Quantitative imaging with the positron emitter (86)Y is the method of choice to determine the uptake and dosimetry of (90)Y-labelled radiopharmaceuticals. To examine the quantitative accuracy of positron emission tomography findings with (86)Y, this non-pure positron emitter was evaluated in a cylindrical phantom with rods of Teflon, water and air and measured with three different scanners: ECAT EXACT (2D/3D), ECAT HR+ (2D/3D) and PC4096+ (2D). After standard reconstruction, (86)Y radioactivity measured with the ECAT EXACT and related to the true radioactivity varied between 0.84 and 0.99 in 2D and between 0.93 and 1.20 in 3D from the first to the last acquisition (eight half-life times later). The water and Teflon rods exhibited considerable amounts of reconstructed radioactivity-21% in 2D and 67% in 3D for water and 65% and 147%, respectively, for Teflon-compared with the actual (86)Y radioactivity of the phantom. For the ECAT HR+ similar results were obtained in 3D, but there were even greater overestimations in 2D. Measurements with the PC4096+ showed rather small errors, with 10% for water and 20% for Teflon. To correct for the background of gamma-coincidences, sinograms were analysed and an experimental percentage of the background was subtracted from the sinograms. In order to minimise the errors in reconstructed radioactivity, the subtraction value had to be different for the individual scanners and modes. Our results demonstrate that (90)Y/(86)Y-based dosimetry for bone and red marrow must be regarded with caution if it is derived from regions of interest over the bone, the density of which is similar to that of Teflon. To obtain more reliable estimates, an appropriate background correction must be applied and tailored individually with respect to the scanner and acquisition mode.