Adaptive algorithms which provide desired performance and have small computational load are key factors that enable the integration of smart antennas into 3G wireless communication systems. We evaluate non-blind (LMS, RLS), blind (CM), and semi-blind (CFA) type receive beamforming algorithms in terms of their performance, complexity, and convergence rate for cdma2000 systems. Simulations are performed for a 5-element uniform linear array operating in a wireless vector channel that consists of multipaths and multiple access interference. Results show that CFA has the highest SINR with minimum deviation in varying fading conditions. CM has the smallest SINR performance possibly due to not involving pilot information, but it has the computation load as low as LMS does, i.e. O(M) where M is the antenna number. RLS achieves SINR comparable to CFA with the fastest convergence rate.