In SPECT imaging of pure beta emitters, such as (90)Y, the acquired spectrum is very complex, which increases the demands on the imaging protocol and the reconstruction. In this work, we have evaluated the quantitative accuracy of bremsstrahlung SPECT with focus on the reconstruction algorithm including model-based attenuation, scatter and collimator-detector response (CDR) compensations. The scatter and CDR compensation methods require pre-calculated point-spread functions, which were generated with the SIMIND MC program. The SIMIND program is dedicated for simulation of scintillation camera imaging and only handles photons. The aim of this work was therefore twofold. The first aim was to implement simulation of bremsstrahlung imaging into the SIMIND code and to validate simulations against experimental measurements. The second was to investigate the quality of bremsstrahlung SPECT imaging and to evaluate the possibility of quantifying the activity in differently shaped sources. In addition, a feasibility test was performed on a patient that underwent treatment with (90)Y-Ibritumomab tiuxetan (Zevalin). The MCNPX MC program was used to generate bremsstrahlung photon spectra which were used as source input in the SIMIND program. The obtained bremsstrahlung spectra were separately validated by experimental measurement using a HPGe detector. Validation of the SIMIND generated images was done by a comparison to gamma camera measurements of a syringe containing (90)Y. Results showed a slight deviation between simulations and measurements in image regions outside the source, but the agreement was sufficient for the purpose of generating scatter and CDR kernels. For the bremsstrahlung SPECT experiment, the RSD torso phantom with (90)Y in the liver insert was measured with and without background activities. Projection data were obtained using a GE VH/Hawkeye system. Image reconstruction was performed by using the OSEM algorithm with and without different combinations of model-based attenuation, scatter and CDR compensations. The reconstructed images were then evaluated in terms of the accuracy of the total activity estimate in the liver insert. It was found that the activity in a large source such as the liver was estimated with a bias of around -70%, when no compensations were included in the reconstruction, whereas when compensations were included the bias obtained was between -10 and 16%. It is concluded that although the (90)Y bremsstrahlung spectrum is continuous with no pronounced peak and the count rate is low, it is possible to achieve reasonably accurate activity estimates from bremsstrahlung SPECT images if proper compensations are applied in the reconstruction. This conclusion was also confirmed by the patient study.