Two systems for integrated weed control in winter wheat based around the combination of herbicides with cultural control have been investigated and compared with conventional practice in experiments between 1993 and 2001. These systems were (a) an overall spray of a reduced herbicide dose followed by spring tine harrow weeding and (b) the combination of herbicide applied over the crop row with a novel vision guided inter-row hoe. The latter required wheat to be established with a wider (22 cm) inter-row spacing than standard (12·5 cm). Experiments over 10 sites/seasons indicated that this increased spacing could be achieved without yield loss. Trials to measure the accuracy of hoe blade lateral positioning using the vision guidance system indicated that error was normally distributed with standard deviation of 12 mm and a bias that could be set to within 1 cm. This performance could be maintained through the normal hoeing period and the crop row location and tracking techniques were robust to moderate weed infestation. In the absence of weeds neither overall harrowing nor inter-row hoeing affected winter wheat yield, 1000-seed weight or specific weight in 12·5 or 22 cm rows. When combined with inter-row hoeing, manually targeted banded applications of fluazolate, pendimethalin or isoproturon reduced grass weed levels and increased yields over untreated controls, though better results were obtained using overall herbicides. However, improvements would be possible with more accurately targeted herbicide applications and more effective inter-row grass weed control. The implications and costs of using such an integrated system are discussed and requirements for future developments identified.