To evaluate the effect of varying ferrule height and distribution on the fatigue resistance and fracture strength of endodontically treated premolars. Thirty-two extracted human premolars were endodontically treated and prepared to receive all-ceramic crowns. Teeth were randomly divided into two groups. The occluso-cervical height was reduced to 2 mm and 4 mm for the 1st and 2nd groups, respectively. According to the ferrule distribution, samples in each group were randomly divided into two subgroups with inadequate ferrule on 1 or 2 proximal walls, respectively. Coronal tooth structure was built-up to 4 mm using a core build-up material retained by a fiber post. The CEREC system (Sirona) was used to design and mill lithium-disilicate crowns. All posts and crowns were cemented to their respective teeth using self-adhesive universal resin cement. All crowns were subjected to cyclic loading in a universal testing machine (Instron) (20-100N, 100,000 cycles, 20Hz). After fatigue, crowns were loaded to fracture. Fracture load data were analyzed using two-way ANOVA at α = 0.05. All crowns passed the fatigue testing without any signs of fracture or crack. Two-way ANOVA revealed a statistically significant effect of the number of missing walls (P < 0.05) on the fracture load, but not the ferrule height or the interaction terms. Samples with one missing wall showed higher fracture load in comparison to samples with two missing walls. The number of missing walls, not the ferrule height, has a significant effect on the fracture load of the lithium-disilicate crowns restoring endodontically treated premolars.