Two field experiments were carried out in 2000 and 2001 on a sandy-loam soil to compare different options (harrowing or hoeing) for mechanical weed control in string bean (Phaseolus vulgaris L.). The first experiment investigated combinations of four tine adjustments and two treatment intensities (one or two passes) of a spring-tine harrow. The second experiment compared four different hoeing systems: (1) a PTO-powered rotary tiller, (2) a precision hoe with sweep and goose-foot shares, (3) a precision hoe (as before) + torsion weeder, and (4) a precision hoe (as before) + torsion weeder operated with the tines crossed. Both experiments also included a weedy check and a post-emergence herbicide treatment as references. Various mechanical work parameters, weed density by species, weed biomass, and string bean yield were determined for all treatments. The machines used in these experiments clearly showed different working characteristics, and their behaviour changed according to soil conditions. In the harrowing experiment, differences among treatments in weed density and biomass as well as in yield were usually negligible, and probably due to low initial weed presence. In contrast, hoeing was always beneficial in terms of weed control and crop yield. Data from these experiments are not conclusive to allow ranking of the mechanical weeding treatments based on their overall efficacy.