A class of adaptive beamforming algorithms was proposed based on directly minimizing the bit-error rate (BER). The popular linear minimum bit error rate (MBER) stochastic beamforming algorithms suffer from low convergence speed and large training data. In this paper, we investigate the use of adaptive stochastic gradient-Newton normalized MBER algorithms in the design of linear beamformer receivers for multipath Rayleigh fading DS-CDMA systems. Simulation analyses show the enhancements in convergence rate and BER performance of the proposed algorithms over the popular minimum mean squared error (MMSE) and other MBER based algorithms in the cost of complexity.