Flexible- and rigid-walled minirhizotron techniques were compared for estimating root length density of 14- to 28-day-old Pinto bean (Phaseolus vulgaris L.) and spring whet (Triticum aestivum L.) plants in soil boxes under controlled environment conditions at three soil bulk densities (1.3, 1.5 and 1.7 g cm−3). The flexible-tube system consisted of bicycle inner tubes inflated inside augered access holes and removed only when measurements were taken. Rigid tubes were constructed of extruded polybutyrate plastic. In both cases tubes were oriented horizontally. Despite similar root densities for wheat and beans based on measurements obtained from soil cores, root densities estimated from both types of minirhizotron were higher in bean than in wheat in uncompacted soil. Estimates of root density by the flexible tube minirhizotron were more closely correlated with soil core image analysis estimates than were those by the rigid minirhizotron system. At high soil bulk density, rigid tube measurements consistently overestimated actual rooting density of both wheat and bean. The relationship between estimated and actual rooting densities in the case of flexible tube measurements was not significantly influenced by soil bulk density. These findings were consistent with the theory that preferential root growth is induced by gaps at the soil-observation tube interface, inherent in the rigid tube technique, and was accentuated under conditions of high soil strength.