A quantitative structure–toxicity relationship (QSTR) model based on multivariate image analysis was built to predict the toxicity of 30 organic compounds against fathead minnow (Pimephales promelas), a fish species of wide occurrence. The derived multivariate image analysis (MIA) applied to QSTR (MIA-QSTR) model provided good correlation between chemical structure and toxicity, i.e., r 2 of 0.96, q 2 of 0.59, and $$ q^{ 2}_{ 20\% } $$ of 0.61 – the squared correlation coefficients of the linear regression of experimental versus predicted 50% lethal concentration (LC50) plots for the calibration, leave-one-out,and leave-20%-out cross-validation, respectively. The method demonstrated the potential to be used as a tool for prediction of toxicity data in vertebrate animals.