Angle-modulated bat algorithm (AMBA) provides a technique that enables the bat algorithm (BA) developed for continuous problems to operate in binary spaces, which employs an angle modulation technique to generate a bit string corresponding to a binary problem solution and applies the BA to evolve the coefficients of the trigonometric generating function. However, there are some identified limitations of the original AMBA. This paper proposes a new AMBA variant called amplitude AMBA (A-AMBA). The A-AMBA is compared with the AMBA and several other binary heuristic algorithms on 13 classical benchmark functions and 12 zero-one knapsack problems. The obtained results indicate that the performance of this variant is superior to the AMBA in many problem cases.