Root length and diameter distribution are important characteristics to be considered when describing and comparing root systems. Root length and root-diameter distribution may be obtained in two ways: by microscopical measurements, which are laborious, or by computerised analysis, which is fast but sensitive to the scanning protocol. Although scanning protocols vary widely between laboratories, papers rarely report the details. Using two commercially available root analysis software packages (Delta-T Scan and WinRHIZO), we performed a sensitivity analysis of scanning protocols for estimating root length and diameter distribution. The results are shown graphically, rather than numerically, because the interactions between different parameters in the overall scanning protocol are most clearly illustrated by the shape of the curves. The present analysis clearly demonstrated the sensitivity of the two scanning methods with regard to staining period, maximum root density, scanning resolution and transformation threshold. For example, estimating the root-diameter distribution versus measuring root length, puts opposite constraints on the transformation threshold settings. We suggest the following settings for the most sensitive parameters: a staining period of 24 h, a root density of less than 0.5 mm root per mm2 surface, a resolution of 400 dpi and the automatic threshold for WinRHIZO and a brightness of 200 for Delta-T Scan. According to this protocol, comparison of computerised analyses with microscopic measurements showed good total root length and diameter distribution agreement for three contrasting root systems. We suggest to always start with the present protocol when studying other species. If validation indicates that the proposed scanning protocol needs to be modified for other species, the present sensitivity analysis may be used as a guideline for changing the most critical parameters. Similarly, the use of another stain than neutral red may also require modifications of the scanning protocol. In general, a long staining period (e.g., 24 h) is recommended for all stains, as small differences in staining period have the least effect when the root tissue is saturated. To enhance comparability of results in the literature, the staining period, stain, (maximum) root density, scanner resolution and threshold should always be listed when root data obtained by scanning are presented.