Slave‐making ants raid nests of other ant species, capture the developing offspring and rear them to slave workers. Here we compare slave‐making of three formicine slave‐making ants: the facultative Formica subnuda, the obligate Polyergus breviceps, and F. subintegra which previously has been considered facultative but appears to be an obligate slave‐making ant. If F. subintegra is an obligate slavemaker, slave‐making of F. subintegra should differ from that of F. subnuda but closely resemble slave‐making of P. breviceps in the following aspects: (1) Obligate slavemakers are rarer than facultative slavemakers. (2) Slaveless colonies of facultative slavemakers are found, but obligate slavemakers always have slaves. (3) Because obligate slavemakers depend on their slaves, they should have a higher proportion of slaves than facultative slavemakers. (4) Owing to special adaptations obligate slavemakers are able to raid bigger colonies, and hence have bigger slaves than facultative slavemakers. (5) Dufour's gland of F. subintegra should be larger than that of F. subnuda. Per 100 free F. podzolica colonies, the number of P. breviceps and F. subintegra colonies with F. podzolica slaves were 1.3% and 3.9%, respectively, and the number of F. subnuda colonies with F. podzolica 3.7%, and without F. podzolica 7.5%. The proportion of slaves, when present, varied between 1–30% in the colonies of F. subnuda, and between 70–90% in the colonies of the other species. The slaves of F. subnuda were significantly smaller than those of F. subintegra and P. breviceps. The length of F. subnuda's Dufour's gland was one third of the length of F. subintegra's gland. The results show that slave‐making of F. subintegra parallels that of P. breviceps, and contrary to the earlier notion, F. subintegra is an obligate slave‐making ant. We suggest that F. subnuda and F. subintegra represent extreme modes of slave‐making behaviour in the Fo