The established methods after Parikh [M. Parikh, IBM J. Res. Dev. 24, 438 (1980)] allow a dose correction using the ‘‘Two Gaussian Model’’ by considering the parameters α, β, and η. A guaranteed accuracy after development cannot be given for these methods because the development process, depending on resist type, thickness and contrast, is not taken into account. In order to calculate a final guaranteed accuracy considering α, β, η, and the full resist development process, we did a calculation in following steps. First, we calculated the proximity correction just for backscattered electrons by the method of ‘‘simple compensation’’ [V. V. Aristov, A. A. Svintsov, and S. I. Zaitsev, Microelectron. Eng. 11, 641 (1989)]. In the second step, we simulated the proximity effect after development (modeling) with the before corrected dose distribution, but now considering all parameters: α, β, η, thickness H, and contrast γ of positive resist. This leads to a guaranteed accuracy δ (maximum structure deviation) for a given design rule L using the correction method of simple compensation. This guaranteed accuracy can be expressed in dimensionless coordinates δ/α=f(L/α,H/α,η,γ). So the accuracy of the electron lithography in this approach is determined by the beam size, characterized by α. Simple compensation results in the accuracy equal to a fraction of α. A better proximity correction below the guaranteed accuracy is possible by using simple compensation in iteration and by correcting for α inside a small structure frame.