Viability is an important component of seed quality, which could be detained by germinating the seeds. Currently testing the seed viability of rice takes a long time (5-14 days), so it becomes a limiting factor in the seed production process. An alternative method for rapid seed viability detection is using the Near Infrared (NIR) spectra and using artificial neural network (ANN) as a data processing system. This research was aimed to study the use of NIR spectra and ANN to predict the viability of rice seeds. NIR reflectance (1,000-2,500 nm) of a Ciherang rice seed samples (40 grams), was used as the input data to develop the ANN model. A total of 60 samples were subjected to accelerated aging to obtain various levels of germination. The development of ANN models was done through calibration and validation of NIR spectra to the viability parameters. As ANN input, NIR reflectance of seed sample was given pretreatment data such as normalization, first derivative, second derivative, standard normal variate (SNV) and principal component analysis (PCA). The results showed that longer accelerated aging caused a decrease in seed viability. This was also indicated by the decrease in soluble protein and an increase in free fatty acids. The intensity of the NIR absorbance spectra also showed the same in the absorption region of soluble protein and free fatty acids. The best ANN models to predict the germination was 10PC-5-3 ANN with the SNV NIR reflectance used as the input data. Coefisien correlation of the validation was 0.8947, the value of ratio performance deviation was 2.2359 and the standard error performance was 9.9233%. The use of NIR spectra and ANN was potentially useful to perdict the viability of rice seeds more rapidly.