Food processing industries routinely acquire large amount of data, but unfortunately not always meaningful information is extracted. For hydrogenated fats industry, the monitoring of changes in the composition of fatty acids is crucial to guarantee a final product with desired properties. The combination of gas chromatography (GC) with artificial neural networks (ANN) may be an alternative to extract the most relevant information about the chemical composition during the hydrogenation. The self-organizing maps (SOM), an unsupervised ANN, are widely used for exploratory data analysis because can deal with nonlinear relationships and produce an easy interpretation graphical output. Were analyzed two partially hydrogenated fats produced in the hydrogenation reactor of an Agro-industrial Cooperative located in the Central-West region of Parana State (Brazil). Samples were collected, using five-minute intervals, from time zero until the end point of reaction which is specific to each product. The separation, identification and quantification of the fatty acids methyl esters (FAME) were performed by GC and the iodine value (IV) was determined according to AOCS (Cd 1c-85). The FAME content and IV were analyzed by a two-dimensional SOM with ten neurons in each dimension which was trained during five thousand epochs. The SOM was able to cluster the samples according to the IV that ranged from 127 to 60 g I2/100g. According to the maps, the variation of mono-unsaturated fatty acids is associated to the increase of C18:1. Poly-unsaturated are mainly correlated with C18:2 which is completely hydrogenated in the beginning of the reaction. Until the monitored point, it was observed an increase in saturated fatty acids and trans fatty acids associated respectively with C18:0 and C18:1(trans) increase. Besides, it was observed that trans-isomers are more stable than the cis-isomers. The SOM with GC was an efficient tool for monitoring the different stages of the industrial hydrogenation process.