A parallel and a distributed implementation for a very important problem in the searching theory, the optimal binary search tree (BST) problem, is presented and analyzed. Implemented as a VLSI array, the algorithm for building the optimal BST uses O(n ²) processors and has the parallel time complexity O(n). A search is solved in O(log n) time. On a cluster of computers, the binary search tree is organized on two levels: the first level corresponds to the BST of searching intervals and the second level is the level of the BST for effective searching within an interval. A hybrid solution is also considered. The best variant depends on the hypothesis of the searching problem.