A new approach is presented for combinatorial development of polymer nanocomposites with compositional gradients (CGs). The CGs were developed using transient processing conditions in twin screw extrusion with small quantities of expensive nanoscale fillers. Convolution of step input with normalized residence volume distributions (RVDs) was used to establish the processing-structure relationship for the CGs. The normalized RVD was established as a process characteristic independent of processing conditions and measured in situ using an optical probe. The CG determined nondestructively using the new combinatorial approach was validated through comparison with more time-consuming and destructive thermogravimetric analysis. The CG could also be established with relatively inexpensive microscale fillers using the normalized RVD obtained with nanoscale fillers, suggesting that transient effects of the mixing process are independent of the size of the filler. Finally, structure-property relationship of combinatorially developed polymer nanocomposites was established by characterizing their dynamic mechanical behavior (storage modulus, G', and loss modulus, G ''). The dynamic mechanical behavior of the combinatorially developed composites correlated well with the batch-processed ones, indicating that the transient mixing conditions in extrusion do not affect the material properties. (C) 2008 American Institute of Chemical Engineers.