An analysis of the data obtained by Parsons et al. (2006a) on runoff and erosion plots in Arizona did not support the conclusion of the authors that the pattern of sediment yield for these plots first increased as plot length increased but then decreased. The approach adopted by the authors was not appropriate for analysing the data from a small number (10) of rainfall events where data were missing from many of the plots because of technical failures. Also, for a proper analysis of the effect of plot length to be undertaken, a shorter plot has to appear to physically represent the top portion of a longer plot. However, the events were spatially and temporally very localized, as they varied in intensity, duration, total rainfall, time to peak etc between the plots. Under these circumstances, data obtained from these experiments cannot be used to determine the effect of slope length on sediment yield. Copyright © 2008 John Wiley & Sons, Ltd.