This article reports the optimal design and measurement conditions for microfluidic diffusion experiments. For this purpose, we conducted and experimentally validated the model-based Optimal Experimental Design (OED) of a pressure-driven H-cell microchannel for diffusion measurements in liquids. While earlier work employed low Fourier numbers (i.e., low contact times), our results show the existence of an optimal finite Fourier number for which the measurement sensitivity increases drastically. For the optimal contact times (Fo = 0.30 ... 0.39), the accuracy of diffusion measurements is improved by orders of magnitude compared to the previously used sub-optimal low Fourier region (Fo < 0.01). A general design rule for optimal H-cell diffusion experiments is derived. To demonstrate the utility of the optimized microfluidic method, diffusion coefficients are measured for a variety of binary mixtures. We show that H-cell measurements designed and operated according to the identified principles lead to a fast, efficient and cheap method for high-throughput diffusion analysis. (C) 2012 Elsevier Ltd. All rights reserved.