Inherently low thermal conductivities of basic fluids form a primary limitation in high-performance cooling which is an essential requirement for numerous thermal systems and micro-devices. Nanofluids, i.e., dilute suspensions of, say, metal-oxide nanoparticles in a liquid, are a new type of coolants with better heat transfer performances than their pure base fluids alone. Using a new, experimentally validated model for the thermal conductivity of nanofluids, numerical simulations have been executed for alumina-water nanofluid flow with heat transfer between parallel disks. The results indicate that, indeed, nanofluids are promising new coolants when compared to pure water. Specifically, smoother mixture flow fields and temperature distributions can be achieved. More importantly, given a realistic thermal load, the Nusselt number increases with higher nanoparticle volume fraction, smaller nanoparticle diameter, reduced disk-spacing, and, of course, larger inlet Reynolds number, expressed in a novel form as Nu = Nu(Re and Br). Fully-developed flow can be assumed after a critical radial distance, expressed in a correlation R-crit = fct(Re), has been reached and hence analytic solutions provide good approximations. Nanofluids reduce the system's total entropy generation rate while hardly increasing the required pumping power for any given Re-in. Specifically, minimization of total entropy generation allows for operational and geometric system-optimization in terms of S-gen = fct (Re and delta). (C) 2010 Elsevier Ltd. All rights reserved.