Abstract: Vaccines require a period of at least three months for clinical trials, hence a method that can identify elicitation of immune response a few days after the first dose is a necessity. Evolutionary variable selections are modeling approaches for proper manipulation of available data which were used to set up an animal model for classification of time dependent 1HNMR metabolomic profiles and pattern recognition of fluctuations of metabolites in two groups of male rabbits. One group of rabbits was immunized with human red blood cells and the other used as control. Blood was obtained every 48 h from each rabbit for a period of six weeks and the serum monitored for antibodies and metabolites by 1HNMR spectra. Evaluation of data was carried out using orthogonal signal correction followed by principal component analysis and partial least square. A neural network was also set up to predict immunization profiles. A distinct separation in patterns of significant metabolites was obtained between the two groups, just a few days after the first and the second dose. These metabolites were used as targets of neural networks where each sample was used as test, validation and training and their quantitative influence predicted by regression. This model could be used for prediction of immunization in rabbits a few days after the first dose with 96% accuracy. Similar animals and human vaccine trials would assist greatly in reaching early conclusions in advance of the usual two month immunization schedule; resulting in an appreciable saving of cost and time.