Abstarct: In this work, Wavelet Neural Network (WNN) has been used to predict density of Alkanes and Cycloalkanes with 2-19 carbon atoms over a wide range of temperature and pressure. We modeled two wavelet neural networks for density prediction: one for liquidified and compressed natural gas (LNG and CNG (C2-C4)) and another for alkanes with 5-19 carbon atoms and cycloalkanes. At first, WNN model has been constructed for LNG and CNG using temperature and pressure variants and 2 descxriptors baxsed on group contribution method namely the number of Methyl, Methylene. Then a computer program was written in MATLAB 6.1 (Math works) to use wavelet neural network. After training of network, the WNN parameters consist of the number of neurons in hidden laxyer, the learning rate, the momentum and the number of iteration have been optimized simultaneously for this. The capability of the optimized model has been evaluated by plotting experimental values of density versus the prediction and validation sets. The obtained result with an average percentage error for density prediction lower than 0.9%. reveals the capability of the WNN model. In the next step, WNN model has been optimized for density prediction of alkanes (C5-C19 ) and cycloalkanes using 5 descxriptors consisting of temperature, pressure and the number of methyl, methylene and Methyn groups, in similar manner the linear fit hold quiet well the correlation coefficient, R^2≥0.9486. This approach gives density of these compounds with average percentage error less than 1.1%.