In this paper, considering the influence of network traffic flow on the spreading behaviors of epidemics and according to the mean-field theory, we investigate the epidemic immunization strategies in scale-free networks, and propose an improved acquaintance immunization mechanism. Theoretical analysis shows that considering the influence of traffic flow, the random immunization can hardly reduce the spreading velocity of epidemics if the density of vaccinated nodes is small. However, the targeted immunization can sharply depress the epidemic spreading even only a tiny fraction of nodes are vaccinated, and the effects of immunizing the most highly connected nodes and vaccinating the nodes with the largest betweenness are almost the same. We also find that if the network global information is unknown, compared with the classical acquaintance immunization strategy, the strategy proposed in this paper can be used to obtain good immune effect. Numerical simulations confirm the theoretical results.