Automated multiple development (AMD) is a gradient elution technique for high-performance thin-layer chromatography (HPTLC), in which successive runs are performed with decreasing solvent strength and increasing developing distance. A method for obtaining separations or distributions of heavy petroleum products according to their number of aromatic rings has been developed using AMD together with caffeine-impregnated silica gel HPTLC layers and selective detection by fluorescence scanning densitometry. A gradient of dichloromethane-heptane, starting from a composition of 70:30 (v/v) to 65:35 (v/v), in 34 elution steps with a migration of 2 mm per step, over a total migration distance of 68 mm, has been used. Calibration of elution has been carried out using polycyclic aromatic compounds (PACs) with a molecular weight (MW) from 200 to 1000 atomic mass units (amu), which have been separated under the described conditions. Other parameters on AMD separation have also been discussed, such as the effect of mobile and stationary phases. The application of this technique to different heavy petroleum products is discussed.