Surface tension as an important characteristic in much scientific and technological research is a function of liquid materials' physical properties. Thus, it is desirable to have an accurate correlation between effective parameters and surface tension. This study investigates the applicability of artificial neural networks as an efficient tool for the prediction of pure organic compounds' surface tensions for a wide range of temperatures. The experimental data gathered for training and verification of the network are related to a wide variety of materials such as alkanes, alkenes, aromatics, and sulfur, chlorine, fluorine, and nitrogen containing compounds. The most accurate network among several constructed configurations has one hidden layer with 20 neurons. The average absolute deviation percentage obtained for 1048 data points related to 82 compounds is 1.57%. The results demonstrate that the multilayer perceptron network could be an appropriate lookup table for the determination of surface tension as a function of physical properties.