In summer 1992, isoprene emission was measured on intact leaves and branches of Quercus alba (L.) at two heights in a forest canopy. Isoprene emission capacity (measured at 30 °C and a photosynthetic photon flux density of 1000 μmol m−2 s−1) was significantly higher in sun leaves than in shade leaves when expressed on a leaf area basis (51 versus 31 nmol m−2 s−1; P < 0.01). Because leaf mass per unit area (LMA, g m−2) was higher in sun leaves than in shade leaves, emissions of sun and shade leaves expressed on a dry mass basis did not differ significantly (99 versus 89 μg C gDW−1 h−1; P = 0.05). Similar measurements in 1995 were consistent with the 1992 data, but data from leaves in more shaded locations demonstrated that isoprene emission capacity decreased with decreasing growth irradiance, irrespective of units of expression. Isoprene emission capacity in leaves of Q. coccinea Muenchh. and Q. velutina Lam. also declined steeply with canopy depth. Emission capacity, on a dry mass basis, showed no obvious pattern with canopy position in Q. prinus L. There was no difference in the temperature response of sun versus shade leaves of Q. alba, but shade leaves exhibited a greater quantum efficiency and saturated at lower irradiance than sun leaves. Rates of isoprene emission measured on branches of Q. alba were approximately 60% of those measured on individual leaves, as a result of self-shading within branch enclosures. It is recommended that within-canopy variation in isoprene emission capacity be incorporated into regional emission models.