Recently, the study of human brain based on multimodal imaging has become one of the new frontiers of neuroscience research. It provides the opportunity to combine modalities to investigate brain architecture from both functional and structural perspectives. Current analyses typically examine these modalities separately, although multimodal methods are emerging to facilitate joint analyses. We present statistical methods for combining information from data collected from functional and structural neuroimaging. The proposed methods aim to address questions arising from multimodal neuroimaging brain connectivity analysis, including how to derive more reliable and accurate measures of the structural and functional connectivity using the diffusion MRI and fMRI data, and how to combine the multimodality imaging to advance understanding of brain networks. We present statistical modeling approaches for helping address these questions. We evaluate the performance of the proposed methods through simulation studies and also illustrate their applications in real-world multimodal neuroimaging data examples.