Elimination of protozoa or defaunation is known to reduce methane emission by ruminants. However, the long-term efficacy of this practice is not well documented. In this work methane and other end products of fermentation were monitored in five adult, rumen-cannulated wethers that had been kept defaunated for two years (Def1) and successively refaunated (Fau, 12 weeks) and defaunated (Def2) during a 6-month-long experiment. Methane was measured for periods of 4 days at about 4-weekly intervals using the SF6 tracer technique, and rumen content samples were taken regularly throughout the experiment. Refaunation was performed by single administration of a mixed protozoal inoculum (103 cells). A noticeable protozoal population was observed at 12 days after inoculation, which was followed by an exponential increase that peaked at 12 ± 3 × 105 cells/mL rumen fluid at 4 weeks and decreased thereafter to 8 ± 4 × 105 cells/mL at the end of the 12-week Fau period. Production of methane, which was 33 ± 8 L/animal.day in Def1, increased (P < 0.05) up to 42 ± 5 L at the end of the Fau period and dropped back at the end of Def2 to 35 ± 7 L (Def1 v. Def2, P > 0.05). Even though protozoal numbers were comparable to conventional animals 20 days after inoculation, the increase in methane production was observed later, at 8 weeks, suggesting that protozoa are not the only microbial factor favouring methanogenesis and that the ecosystem needs a relatively long adaptation time following protozoal introduction to optimise this function. The presence of protozoa did not affect total volatile fatty acids and acetate concentration in the rumen but increased butyrate, while decreasing propionate concentration (Fau v. Def1, Def2, P < 0.05). These results show that the decrease in methane emissions induced by defaunation in sheep was stable for a period of up to 2 years. In the absence of protozoa, methane decreased by ~20% in both short- and long-term defaunated animals.