Rice cvs with better hardpan penetration would be expected to be more drought resistant in the rainfed lowlands. Although laboratory methods to facilitate the identification or breeding of cvs with good root-penetration ability have been described, there is a need to validate such screens against field performance. Here, we compare previous field measurements with laboratory screening measurements in eight cvs (IR20, CT9993, KDML 105, IR58821, NSG 19, IR62266, Mahsuri and IR52561). These were screened (together with Moroberekan, SG329 and IR36 for comparative purposes) using a flooded wax-layer screen. Of the eight cvs, IR58821 gave the best penetration of a 60% wax layer, with a mean penetration of 5.8 root axes per plant. The worst performer was IR52561, with a mean of 0.6 axes per plant. The cvs IR20, CT9993, KDML 105, IR58821 were also screened (together with Azucena, Bala, Moroberekan, Kinandang Patong and IR36 for comparative purposes) using a (non-flooded) sand-core screen. The sand-core screen allowed mechanical impedance of the whole sand core to be varied independently of aeration and water status. High impedance treatments were obtained by placing weights on the sand cores, which greatly decreased root growth, although differences between cvs in response to impedance in the sand-core screen were small. The ability of rice roots to penetrate wax layers did not appear to be related to their elongation through strong sand, but rather to their ability to resist buckling on encountering the wax layer. Comparison with field measurements showed that cvs with good performance in the wax-layer screen did not necessarily have good hardpan penetration in the field, although IR58821 was the best performer in the field. It is concluded that further work is required to compare root penetration in the field with root penetration in laboratory screens.