Coral gravel soils, which are composite soils consisting of finger-coral fragments and silt matrix, are often found in coastal regions of sub-tropical islands. In this study, for reconstituted soils with various volumetric percentages of coral fragments up to 44% (coral particle fraction of 59%) as the densest package, a series of triaxial CU-bar and CD tests was conducted to study the determination method for the soil design parameters in consideration of the interaction between the soil skeleton, consisting of coral fragments, and the silt matrix. For samples with volumetric percentages of coral fragments less than 20% (coral particle fraction of 31%), the shear strength obtained from the CD tests was slightly larger than that obtained from the CU-bar tests; however, the difference between the two tests was very small. For samples with volumetric percentages of fragments more than 20% (coral particle fraction of 31%), the shear strength obtained from the CU-bar tests was significantly overestimated because of the unrealistically large negative excess pore pressure in the field corresponding to significant dilation. The shear strength obtained from the CD tests also showed a similar tendency corresponding to volume expansion; however, these values are much smaller than those obtained from the CU-bar tests. For the samples with a dense package of coral fragments, shear resistance angle ϕ was much larger than that for normal soils; however, it tended to decrease in association with the particle crush of coral fragments. The tendency of the particle crush was visually evidenced through CT-images taken before and after the triaxial tests. The soil parameters were significantly influenced by the volumetric percentages of the coral fragments in association with particle interaction and particle crush, when the percentage was more than 20% (coral particle fraction of 31%) for the coral gravel soils examined in this study.