We study the inverse problem of quantitative photoacoustic tomography in the framework of Bayesian inverse problems. Posterior distribution and point estimates are examined. Furthermore, we study modelling of errors in situations in which the forward model contains uncertainties.