Early identification and prediction of adverse responses of the environment to the use of chemical and other industrial processes may reduce the enormous cost and wasted opportunities in preserving quality of life. On the other hand, existing experimental tests on organ toxicity, which involve animals and a series of in vivo and in vitro methods are of long duration and extremely high cost. This warrants extensive and intensive search for novel alternative approaches in toxicology that partly or fully avoid the use of experimental testing. We suggest to employ the so-called electron-conformational method of computer-based pharmacophore identification and bioactivity prediction in drug design, worked out earlier [1], in order to evaluate the so-called toxicophore and predict toxicity. In this method, a computational chemistry approach is used to analyze the existing data on specific organ toxicity of different compounds in order to reveal the special group of atoms in these molecules, their space conformation and electronic structure that is responsible for the toxicity (the toxicophore). To do this, the conformational and electronic structure parameters of a set of compounds with known activity (inactivity) is calculated and arranged in special tables (matrices), which are then compared by means of a special computer program to reveal the matrix elements that are common for the active compounds and absent in the same combination in the inactive ones. In this way the conformational and electronic parameters of activity, the pharmacophore or toxicophore, are estimated. Further parameterization of toxic compounds in comparison with similar non-toxic ones allows us to use statistical method to deduce a formula that predicts the toxicity quantitatively. The method has a high prediction power with over 90% probability and reveals also the relative importance of different parts of the molecule in producing the toxicity thus allowing for experimental control of the toxicity by means of purposeful substitutions. The method is worked out in a package of user-friendly computer programs.