The TCR repertoire mobilised against any given peptide (p) pMHC antigen can be either unpredictable (diverse) or predictable (biased). Likewise, the number of clonotypes mobilised against a pMHC antigen can swing considerably from one (monoclonal) to dozens (polyclonal). The mechanism behind this curious biological dichotomy has been a mystery for over 25 years. In an attempt to deconstruct the workings of this central immunological system we performed the largest epitope-specific TCR metastudy ever undertaken. Using a combination of published and newly generated TCRβ sequence data we built a database comprising 38,614 TCR TCRβ sequences across 570 samples for 76 HLA-restricted epitope-specific T cell responses. Multiparametric analysis revealed that the length of the peptide in the HLA groove could directly predict both the clonotype diversity and TCR gene make-up of an epitope-specific response. Additionally, by merging the TCR sequence database with a secondary database comprising 33 published and newly generated pHLA crystal structures we were able to directly link structural parameters with TCR repertoire formation. Specifically, we found that the number of bonds between the peptide and MHC was directly associated with the number of clonotypes mobilised. Additionally, we found that the surface complementarity score between the peptide and MHC was directly associated with the number of TRBV genes mobilised. This data brings us one step closer to predicting and manipulating the TCR repertoire for the purposes of rational vaccine design and therapeutic intervention.