ABSTRACT. The quality of optoacoustic tomographic reconstructions can be severely affected by acoustic reflections or scattering arising at interfaces of highly mismatched organs, such as bones, lungs, or other air-containing cavities. We present a procedure to reduce the associated artefacts based on estimation of the acoustic scatterers distribution within the imaged object. Signals generated by a strong optical absorber are processed and used in a weighted back-projection algorithm. Experimental results in a tissue-mimicking phantom clearly demonstrate improved performance as compared to the case in which no information on the distribution of acoustic scatterers is available.