Correlation between nodes is found to be a common and important property in many complex networks. Here we investigate degree correlations of the Barabasi-Albert (BA) scale-free model with both analytical results and simulations,and find two neighbouring regions, a disassortative one for low degrees and a neutral one for high degrees. The average degree of the neighbours of a randomly picked node is expected to diverge in the limit of infinite network size. As a generalization of the concept of correlation, we also study the correlations of other scalar properties, including age and clustering coefficient. Finally we propose a correlation measurement in bipartite networks.