Mid-range flare (MRF) of an ArF exposure tool induces resist critical dimension (CD) variation depending on local mask (or resist) pattern density. CD correction techniques such as mask CD modification are needed to compensate MRF-induced CD variation and obtain sufficient resist CD uniformity. For this purpose, MRF should be accurately characterized and distinguished from other factors of similar influence caused by photoresist, mask, and so on. We have investigated a method of measuring magnitude and affected range of MRF in an exposure tool easily. In this method, double exposure of L/S with large area and a small square "window" in a widely shielded area is executed on a positive photoresist layer. After bake and development, resist CDs of the L/S are measured from near the window to far from the window using scanning electron microscope. An overcoat layer is used to exclude the influence of acid evaporation from photoresist and re-sticking. L/S mask shows small CD variation because local pattern density is uniform, and so influence of mask error on MRF measurement is small. Influence of MRF is shown in the graph of distance from window edge versus CD. From the distance-ΔCD curve, point spread function (PSF) which represents the characteristics of MRF can be obtained. Comparison of experimentally obtained MRF-induced CD variation, which appears in the periphery of a memory cell area, and calculated CD variation using the PSF obtained in the exposure tool showed good agreement.