Skin serves as a first target of the external fields, since its connective tissue content is high which reflects the biochemical and physiological conditions of the skin and has a positively charged collagen molecule. This study has two main purposes: firstly, it was planned to assess whether the skin of guinea pigs was affected by magnetic fields (B) by determining the collagen synthesis in the skin exposed to 50 Hz magnetic fields of 1 mT, 2 mT and 3 mT with the periods of 4 hours/day and 8 hours/day for 5 days and secondly, it was aimed to model this effect on hydroxyproline concentrations in the skin using Neural Networks. One of the important tasks regarding these types of studies is the modeling of the effect for further use without waste of animal which will form as a database for researchers. In this sense, Neural Networks have been used to serve as a robust tool for the modeling of complex relationships that exists between dependent and independent variables where this relationship can not be effectively modeled by conventional regression methods. A novel approach for the selection of optimal Neural Network architecture has been used. The accuracy of the proposed NN model is defined by standard deviation and correlation coefficient which is found to be quite high (R=0.98). Thus parametric studies are performed to see the influence of each parameter by using the proposed NN model. The results of the study are very promising as it will serve as a data base for researchers in these kinds of studies.