Quantitative structure-toxicity relationship (QSTR) studies on toxicity of 91 organic compounds to Chlorella vulgaris have been performed using extended topochemical atom (ETA) indices using different statistical tools like stepwise regression analysis, multiple linear regression with factor analysis (FA-MLR) as the preprocessing step, partial least squares (PLS) regression and principal component regression analysis (PCRA). The ETA models have been compared with the non-ETA ones derived from different topological and physicochemical parameters The results show that the QSTR models using ETA descriptors (FA-MLR: Q(2)=0.832, PLS: Q(2)=0.891, stepwise: Q(2)=0.867, PCRA: Q(2)=0.741) are comparable to the non-ETA models (FA-MLR: Q(2)=0.794, PLS: Q(2)=0.897, stepwise: Q(2)=0.907, PCRA: Q(2)=0.772). Improved results are obtained (except in case of PCRA) when we considered ETA and non-ETA descriptors in combination (FA-MLR: Q(2)=0.850, PLS: Q(2)=0.913, stepwise: Q(2)=0.909, PCRA: Q(2)=0.671). The best two models are obtained using combined ETA and non-ETA descriptors applying stepwise regression [Q(2)=0.909] and PLS [Q(2)=0.913] techniques. The statistical quality of the best models is better than that of the previously published models. The results suggest that the ETA descriptors are sufficiently rich in chemical information and have ability to encode the structural features contributing significantly to the toxicity of organic chemicals to C. vulgaris.