In　current　paper,　a　quantitative　structure-Activity　relationship　(QSAR)　study　was　performed　for　the　prediction　of　acute　toxicity　of　aromatic　amines.　A　set　of　56　compounds　was　randomly　divided　into　a　training　set　of　46　compounds　and　a　test　set　of　10　compounds.　The　electronic　and　topological　descriptors　computed　by　the　Scigress　package　and　Dragon　software　were　used　as　predictor　variables.　Multiple　linear　regression　(MLR)　and　support　vector　machine　(SVM)　were　utilized　to　build　the　linear　and　nonlinear　QSAR　models,　respectively.　The　obtained　models　with　five　descriptors　show　strong　predictive　ability.　The　linear　model　fits　the　training　set　with　R2　=　0.71,　with　higher　SVM　values　of　R2　=　0.77.　The　validation　results　obtained　from　the　test　set　indicate　that　the　SVM　model　is　comparable　or　superior　to　that　obtained　by　MLR,　both　in　terms　of　prediction　ability　and　robustness.