A method for counting root intersections with observation tubes (mini- or micro-rhizotrons) is proposed that allows a theoretical conversion of root counts to estimated root length density and which is robust to the effects of tracking along the tubes. A field test showed that the method agreed well with measured root length densities in cores for wild cherry roots but not for roots of pasture species. The circumstances in which the method might be expected to perform well are discussed. Analysis of data from a field trial revealed that root counts were not uniformly distributed around the circumference of the observation tube. Implications for the design of such trials are noted.