Exciton diffusion in organic materials is crucial for designing and optimizing organic photovoltaic devices. This paper addresses the important complication of the choice of film thickness affecting measurement accuracy in the method of measuring exciton diffusion length from thickness-dependent photocurrent response in organic bilayer heterostructures. Based on the photocurrent model integrated with optical interference effect, the correlation between the exciton diffusion length and the optimal thickness of organic layer is analytically solved. It shows that the intrinsic nonlinearity of the correlation causes prominent uncertainty of estimated diffusion length. Two approaches were proposed to improve the measurement accuracy: (1) the thickness effect of each layer in this heterostructure on photocurrent response is asymmetric so that certain thickness combination can be chosen to reduce the uncertainty; (2) on account of the limited accuracy of the optimal thickness determined by photocurrent spectra, estimated diffusion length should be verified by spectrum fitting. Taking these improvements into account, the exciton diffusion length in copper phthalocyanine film is determined as 20 +/- 5 nm from curve fitting by this method. (c) 2008 Elsevier B.V. All rights reserved.