Cetane number improvement of diesel fuels is a difficult task that refiners will face in the near future. Aromatics saturation by deep hydrogenation is a necessary, but perhaps not sufficient step in the diesel treatment. Some researchers have proposed selective ring opening (SRO) as an additional step in the upgrading. In this work, we explore some possible reaction pathways of compounds typically found in diesel after different levels of hydrogenation, i.e. decalin (decahydronaphthalene), perhydrophenanthrene, tetralin (1,2,3,4-tetrahydronapthalene), as well as 1-ring and 2-ring aromatic phenanthrenes. We have estimated the cetane number (CN) of each individual compound involved in the reaction pathways, using an artificial neural network program that was trained with pure compound cetane numbers from a database. The results demonstrate the great challenge that reaching high CN represents. In the conversion of decalin, acidic catalysts alone are not able to yield products with CN significantly higher than the decalin feed. Similarly, no significant gain in CN can be expected with hydrogenolysis metal catalysts operating via the dicarbene mechanism. Only in the case of selective metal-catalyzed hydrogenolysis, with preferential cleavage at substituted C-C bonds, the predicted products have CN substantially higher than the decalin feed. As expected, branching has a strongly negative effect on the CN and it should be minimized. Both, metal-catalyzed di-carbenium C-C cleavage and acid-catalyzed ring contraction/ring opening combination leave branching groups in the product. Similarly, the acid-catalyzed ring opening of perhydrophenanthrene does not result in a significantly higher CN than the initial feed. The possibility of minimizing hydrogen consumption in the CN improvement process by an initial partial hydrogenation followed by ring opening was tested by using phenanthrene and tetralin as probe molecules. In the first reaction strategy, partially hydrogenated phenanthrenes (1-ring and 2-ring aromatics) were followed by ring opening of one of the saturated rings. Although this option would lead to lower overall hydrogen consumption, it results in products of much lower CNs than the ones obtained by full hydrogenation of phenanthrene. Similar results are obtained for tetralin. From this analysis, it is clear that upgrading CN of diesel requires extensive hydrogen consumption. For further upgrading, highly selective hydrogenolysis catalysts are needed in order to minimize branching and therefore obtain high CN products. (C) 2005 Elsevier Ltd. All rights reserved.