The variability in amino acid axial rise per residue of the collagen helix is a potentially important parameter that is missing in many structural models of fibrillar collagen to date. The significance of this variability has been supported by evidence from collagen axial structures determined by electron microscopy and X-ray diffraction, as well as studies of the local sequence-dependent conformation of the collagen helix. Here, sequence-dependent variation of the axial rise per residue was used to improve the fit between simulated diffraction patterns derived from model structures of the axially projected microfibrillar structure and the observed X-ray diffraction pattern from hydrated rat tail tendon. Structural models were adjusted using a genetic algorithm that allowed a wide range of structures to be tested efficiently. The results show that variation of the axial rise per residue could reduce the difference metric between model and observed data by up to 50%, indicating that such a variable is a necessary part of fibril model structure building. The variation in amino acid translation was also found to be influenced by the number of proline and hydroxyproline residues in the triple helix structure.