Microfluidic technology enables the creation of well-defined cell culture environments, which integrate the control of multiple biophysical and biochemical cues for designing novel in vitro assays. Growth-factor concentration gradients play a critical role in a wide range of biological processes ranging from development to cancer, guiding cell migration and influencing cell signaling. We present a microfluidic device capable of generating stable concentration gradients in a 3D matrix, while allowing for direct imaging of cellular behavior. The design consists of polydimethylsiloxane microchannels interconnected through 3D matrices. Optical access of the 3D matrix permits direct observation of invasive properties of cells seeded inside the channels or embedded in the matrix. An important characteristic of the microfluidic platform is the capability to generate reproducible, stable and quantifiable concentration gradients that are essential for systematic studies of soluble factor signaling in chemotaxis assays. To characterize the concentration gradients in the device we combine intensity measurements using fluorescent markers and a finite element model, while we also measure the hydraulic permeability and diffusion coefficient of a 40 kDa chemoattractant across the 3D matrix. The numerical model solves the coupled convection-diffusion-Brinkmann equations using a commercial finite element solver. Comparison of measured and computed concentration profiles demonstrates very good agreement, while the simulation can be used as a tool for optimizing the microfluidic design. To demonstrate the device capabilities and the effects of concentration gradients on cell migration, we seeded a brain cancer cell line (U87MG) on the microfluidic channels and monitored cell invasion in 3D type I collagen matrices under control and epidermal growth factor (EGF) gradient conditions. We find that in the presence of an EGF concentration gradient tumor cells are guided towards higher EGF levels and that the directional bias is dependent on the gradient magnitude and EGF concentration. This assay provides new data describing cancer cell invasion in a 3D matrix under control of EGF concentration gradients.