We present infrared photometry of the WC8 Wolf–Rayet system WR 48a observed with telescopes at the European Southern Observatory, the South African Astronomical Observatory and the Anglo Australian Telescope between 1982 and 2011, which show a slow decline in dust emission from the previously reported outburst in 1978–79 until about 1997, when significant dust emission was still evident. This was followed by a slow rise, accelerating to reach and overtake the first (1978) photometry, demonstrating that the outburst observed in 1978–79 was not an isolated event, but that they recur at intervals of 32+ years. This suggests that WR 48a is a long-period dust maker and colliding-wind binary. The locus of WR 48a in the (H−L), K colour–magnitude diagram implies that the rate of dust formation fell between 1979 and about 1997 and then increased steadily until 2011. Superimposed on the long-term variation are secondary (‘mini’) eruptions in (at least) 1990, 1994, 1997, 1999 and 2004, characteristic of relatively brief episodes of additional dust formation. Spectra show evidence for an Oe or Be companion to the WC8 star, supporting the suggestion that WR 48a is a binary system and indicating a system luminosity consistent with the association of WR 48a and the young star clusters Danks 1 and Danks 2. The range of dust formation suggests that these stars are in an elliptical orbit having e∼ 0.6. The size of the orbit implied by the minimum period, together with the WC wind velocity and likely mass-loss rate, implies that the post-shock WC wind is adiabatic throughout the orbit – at odds with the observed dust formation. A similar conflict is observed in the ‘pinwheel’ dust-maker WR 112.