Field tests of a prototype microwave-based weed killer machine were conducted on Abutilon theophrasti, Panicum miliaceum, lucerne and oilseed rape pure stands. The approach can be considered a thermal weed control method, the microwave radiation causing dielectric heating of plant tissue water that eventually kills the plant. The method could overcome the limitations of other thermal methods, such as fire risk with flaming or the heavy loads required for hot water treatments. Species were effectively controlled by microwave irradiation, but their sensitivity and the evolution of damage symptoms over time differed. Lucerne showed no sigmoidal response and was the least affected by the treatment, while a log-logistic curve expressed the dose–response relationships of the other species quite well. The estimated microwave dose for a 90% dry weight reduction ranged from 1015 kJ m−2 in A. theophrasti to 3433 kJ m−2 in P. miliaceum. Energy cost evaluation indicated that increased efficiency is required for this technique to compete with other thermal methods. Microwave efficiency could be increased by a flux configuration that minimizes soil penetration and maximizes absorption by plants, which, in turn, depends on plant growth form.