AbstractAccession, decomposition and accumulation of litter were studied in three sub‐alpine eucalypt forest communities (dominated by overstoreys ofEucalyptus delegatensis, E. paucifloraorE. dives) located in the Brindabella Range. Australian Capital Territory, at an elevation of 1100–1250 m. The sites had either been protected from fire for more than 20 years or been burnt by low‐intensity prescribed fires. After a prescribed burn, the rate of decomposition of abscised leaves was reduced by 22% inE. delegatensisforest and by 34% inE. paucifloraforest, but was little affected in the drierE. divescommunity. Lowered decomposition was apparently due to greater aridity after fire, a consequence of removal of the shading understorey and reduction in the depth and hence mulching effect of the titter layer. Litter accumulates rapidly after prescribed burning, reaching a mass of 10–12 t ha−1within 4–5 years in all communities. Such quantities are dangerous from a fire control viewpoint. The quasi steady‐state mass of accumulated litter ranges from about 17 t ha−1in E. divesandE. pauciflora forests to about 25 t ha−1in old‐growthE. delegatensisforests. The rapid re‐accumulation of litter after fire is not the result of any significant change in litterfall rate, but is due to a marked reduction in the total amount of litter decomposing—and this reduction is more a consequence of a decrease in the weight of the forest floor than to any fire‐induced lowering of the rate of litter decomposition. The rapid build‐up of litter is a consequence of the relatively high rates of litterfall (3.4–5.0 t ha−1year−1) and low rates of litter decomposition (k = 0.19–0.32 year−1) in these forests. In most cases the pattern of litter accumulation was well described by an exponential equation of the form Xt= Xss(1—e‐kt), where Xtis the weight (t ha−1) of litter accumulated at time t (year). Xssis the weight of litter accumulated under steady‐state conditions, and k is a decomposition rate constant (year−1). Marked temporal variations in annual litterfall and mass of accumulated litter were found at specific forest sites which had been unburnt for more than 4.5 years. Variation from the long‐term mean was greater for litterfall (31–37%) than for accumulated litter (14–26%). The maximum error when calculating decomposition rate (k) as the ratio of annual litterfall: accumulated titter, when based on single measurements of these parameters, ranged from 43 to 69% of that based on long‐term measurements. Decomposition rates of the entire titter layer, calculated for periods of 22–79 months, and based on measurements of litter input and change in mass of accumulated titter, were positively correlated with the average number of days per month during each period that the litter layer remained moist (>approx. 60% ODW). The implications of these findings for fire management planning in su