The majority of known genomic information has unknown functional classification. The classification of such genetic elements is an active research topic in the field of bioinformatics. One commonly utilized method for predictive classification of high-dimensional datasets is the random forest (RF) machine learning algorithm. Developed in the early 2000's by Leo Breiman, the RF algorithm uses an ensemble of (for our application) classification trees grown with a randomly selected subspace of the predictor variables to make predictive classifications of a dataset. In this project, we utilize Service de Bioinformatique des G_nomes et des R_seaux's (BiGRe) A CLAssification of Mobile genetic Elements (ACLAME) database as our training observations to design a RF method for classification of major capsid proteins given an amino acid string. The various parameters used to construct the RF ensemble classifier, and the methods for deciding those values, are explored to obtain an algorithm with an estimated type 1 error of 7% and estimated type 2 error of 3.5%. A labeled subset of NCBI's Reference Sequence Database (RefSeq) is used to test the performance of our RF classifier.