Purpose: The length of Descemet's membrane and donor graft sizes in deep anterior lamellar keratoplasty do not match in very steep corneas, which can lead to Descemet's membrane folds. The aim of this study is to establish a theoretical model for graft size calculations for deep anterior lamellar keratoplasty and evaluate its efficacy for preventing Descemet's membrane folds. Methods: We calculated the arc diameter of the recipient bed by using the cosine formula and developed a table to aid surgeons in donor punch size selection. To test the usefulness of this formula, we evaluated the development of Descemet's membrane folds in keratoconus patients with very steep corneas (K >60 D). In group 1, deep anterior lamellar keratoplasty surgeries were performed using graft sizes that were determined based on our model (n=31). In group 2, graft sizes were determined based on the empirical judgment of the surgeon without any formal calculation (n=30). Results: Our theoretical calculations demonstrated that the diameter of donor punch sizes needed to prevent Descemet's membrane fold increases when the cornea is steeper, or the trephine size is larger. We tested the efficacy of this model on the clinical outcome of deep anterior lamellar keratoplasty. The mean age (28.9 +/- 10.1 years vs. 32.8 +/- 8.3 years, p=0.11) and preoperative K1 (59.2 +/- 9.3 D vs. 58.1 +/- 9.4 D, p=0.67), K2 (66.2 +/- 6.0 D vs. 65.7 +/- 7.4 D, p=0.81), and Km values (62.1 +/- 7.7 D vs. 61.8 +/- 8.1 D, p=0.88) were similar between the two groups. Three patients developed Descemet's membrane folds in group 2, and none of the patients developed Descemet's membrane folds in group 1. These results supported our theoretical calculations. Conclusion: Adjustment of donor graft size based on the calculated arc diameter of the recipient bed reduced the development of Descemet's membrane folds after deep anterior lamellar keratoplasty in steep corneas.