To analyze the topological properties of Beijing public transport network, until July 2010, we have collected 1165 bus lines and 9618 bus stops of Beijing City (14 districts and 2 counties) as the sample data to build up a directed and weighted complex network model based on neighboring stops by applying the complex network theory. In this model, bus stops are considered as nodes of the complex network, while bus lines connecting two neighboring stops as edges. Consequently, the network has the topological properties of a complex network and meanwhile the nodes (bus stops) have clear geographic coordinates. The complexity of Beijing public transport is then verified through analyzing the topological properties of node degree, node strength, strength distribution, average shortest path, clustering coefficient in the complex network. We find that the distributions of node degrees and node strengths are extremely uneven and the cumulative strength distributions of the top 5% and 10% nodes reach 22.43% and 43.02% respectively. The results also show that the node strength, ordinal number and cumulative strength distribution of the nodes all follow the power-law distribution, showing the network characteristics of scale-free and small world. Some "key nodes" play an important role in network connection. We find two kinds of "key nodes" by using high carrying pressure node analysis and extract regional central analysis. These rules provide new references for optimizing the urban transport network, managing traffic congestion and planning and developing the traffic.