The Multidimensional knapsack problem (MKP) is a well-known optimization problem with which many real-world engineering problems can be modeled. Due to its NP-hard nature, exact methods of solving the MKP are limited to small-scale problems. Therefore, in the past decade, various meta-heuristics have been developed to solve the MKP in a reasonable time. Butterfly optimization algorithm (BOA) is a recently developed meta-heuristic that has shown good convergence ability as well as avoiding local optima stagnation. In this paper, a binary version of BOA (BBOA) is proposed to solve the 0-1 MKP. BOA is originally designed for a continuous search space therefore in this paper, we propose six binary versions of BOA using three S-shaped and three V-shaped transfer functions and determine the most effective version through experiments. The proposed BBOA also includes an initial population generator and a repair operator based on the pseudo-utility. To evaluate the proposed method, 11 medium-scale and large-scale benchmark problems are employed and BBOA is compared to other competitive algorithms.