The importance of several leaf constituents in controlling rates of litter breakdown was assessed, a pivotal ecosystem process in forested headwater streams. Leaves of 26 tree or shrub species were assembled into monospecific packs (5 g dry mass), enclosed in coarse (1 cm) and fine mesh (0.5 mm) bags and exposed in a third-order mountain stream. Four replicate bags were retrieved after 20 days, 40 days or both, and analysed for ash-free dry mass (A.FDW) remaining. Exponential decay rates of litter ranged from 0.003 to 0.047 day-1, corresponding to estimated halflives of 15-210 days. Four separaie packs of each leaf species not exposed in the stream were analysed for initial concentrations of nitrogen (O.l-2.4% of AFDW), phosphorus (0.01-0.30% of AFDW), cellulose (13-32% of AFDW), and proximate lignin ( 4-38% of AFDW). Regression analyses showed that contrary to common belief and results from terrestrial studies. breakdown rate was not related to the initial N content of litter (r2 = 0.23 and r2 = 0.06, coarse and fine mesh, respectively). Likewise, no significant relationship was found with the concentration of P (r2 = 0.08 and r2 = 0.07, coarse and fine mesh, respectively). However, the initial lignin content ofleaves showed a highly significant curvilinear relationship with breakdown rate, explaining 73% of the total variance in coarse and 82% in fine mesh bags. Breakdown rate was also significantly related lo cellulose content, but this efiect appeared to be due to a partial co-vmiation of lignin and cellulose contents. Other combined parameters such as the widely used lignin:N ratio also showed weaker relationships. These results suggest that litter breakdown was strongly governed by litter carbon quality, thus differing fundamentally from the typical process in forest soils. The apparent lack of nutrient control may be due to the abundant supply of N and P in stream water flowing across the decaying leaves, which is unavailable in terrestrial systems.