[en] Tissue perfusion is frequently determined from dynamic contrast-enhanced CT or MRI image series by means of the steepest slope method. It was thus the aim of this study to systematically evaluate the reliability of this analysis method on the basis of simulated tissue curves. 9600 tissue curves were simulated for four noise levels, three sampling intervals and a wide range of physiological parameters using an axially distributed reference model and subsequently analysed by the steepest slope method. Perfusion is systematically underestimated with errors becoming larger with increasing perfusion and decreasing intravascular volume. For curves sampled after rapid contrast injection with a temporal resolution of 0.72 s, the bias was less than 23% when the mean residence time of tracer molecules in the intravascular distribution space was greater than 6 s. Increasing the sampling interval and the noise level substantially reduces the accuracy and precision of estimates, respectively. The steepest slope method allows absolute quantification of tissue perfusion in a computationally simple and numerically robust manner. The achievable degree of accuracy and precision is considered to be adequate for most clinical applications. (orig.)