In the laboratory, the trail pheromone methyl 4-methylpyrrole-2-carboxylate (M4MP2C) when added to bait particles acts as an attractant to leaf-cutting ants, increasing the pick-up of pheromone-impregnated bait compared with plain bait. However, field trials in Trinidad, Brazil and Paraguay were disappointing. Bait with and without pheromone was scattered in the foraging area, and the proportion of each taken to the nest by Atta sexdens rubropilosa Forel, A. cephalotes (L.) and Acromyrmex octospinosus (Reich) was assessed. In only one of the experiments (A. octospinosus foraging bait with 50 p.p.m. of M4MP2C) was increased pick-up detected. Four reasons for this are suggested: (1) small pheromone effects are difficult to measure as scattering bait produced a patchy distribution and heterogeneous replicates, whilst choice trials on plain and pheromone-impregnated bait placed close together were difficult to interpret, as the pheromone probably excited the ants, increasing pick-up of plain bait; (2) A. octospinosus workers, the subject of detailed studies, were less sensitive to M4MP2C in the field than in the laboratory; (3) as the pheromone only attracts, it does not produce a comparable increase in the number of pieces picked up, contact/pick-up ratios of sucroseimpregnated discs increasing from 2·26 at 5 pg pheromone to 16·39 at 50 ng; and (4) soyabean oil, citrus-pulp extract and orange juice were found to be attractive, and when pheromone was added, its effects were not additive. Early studies used sucrose-impregnated paper discs, and as these have no attractants of their own, the addition of M4MP2C had a more marked effect. It is concluded that M4MP2C is not a cost-effective addition to current leaf-cutting ant baits which possess food odours, a much cheaper source of attractiveness.