Prostaglandins (PGs) promote both bone resorption and formation in vitro and in vivo. In a synchronised model of bone remodelling, indomethacin, an inhibitor of PG synthesis, given from the start of the sequence, transiently impaired bone resorption. In this study we further explored the involvement of PGs in this model by treating rats with indomethacin (7.5 mg kg-1 day-1) for 6 days from the peak of resorption (day 4 after activation in this model) or during reversal (day 6 after activation). In rats treated from day 4, the resorption surface (Oc.S/BS) and the number of osteoclasts (N.Oc/BPm) were higher on day 10 (+69 %, P < 0.01, and +60 %, P < 0.02 compared with controls, respectively); no effect on cell resorptive activity was observed. The bone formation surface (OS/BS) was reduced (-50 %, P < 0.01). The inactive surface (In/BS) was not modified. In rats treated from day 6, the Oc.S/BS was also higher than in controls (P < 0.02), as was the N.Oc/BPm (P < 0.05). Osteoclast activity appeared to be increased, as the osteoclast-bone interface was larger (P < 0.02), but the mean lacuna area was reduced (-23 %, P < 0.05). Bone formation was also strongly affected: the OS/BS was decreased (-66 %, P < 0.01), as was the osteoid seam thickness (-24 %, P < 0.05). The In/BS was increased 1.5-fold (P < 0.05). These data indicate that PGs intervene at various stages of this remodelling sequence, as both resorption and formation were affected by indomethacin. Although resorption resumed in the two treatment groups despite treatment continuation, the timing of treatment was clearly important. Only inhibition of PG synthesis at the peak of resorption delayed all phases of the remodelling sequence. In contrast, inhibition during the reversal phase prevented activation of a significant part of the bone surface usually involved at this stage of remodelling; this treatment schedule reduced the resorptive capacity of the system, and depressed osteoblast activity. Experimental Physiology (2001) 86.3, 373-379.