Abstract. Isoprene emissions are a key component in biosphere–atmosphere interactions, and the most significant global source is the Amazon rainforest. However, intra- and interannual variations in biological and environmental factors that regulate isoprene emission from Amazonia are not well understood and, thereby, are poorly represented in models. Here, with datasets covering several years of measurements at the Amazon Tall Tower Observatory (ATTO) in central Amazonia, Brazil, we (1) quantified canopy profiles of isoprene mixing ratios across seasons of normal and anomalous years and related them to the main drivers of isoprene emission – solar radiation, temperature, and leaf phenology; (2) evaluated the effect of leaf age on the magnitude of the isoprene emission factor (Es) from different tree species and scaled up to canopy with intra- and interannual leaf age distribution derived by a phenocam; and (3) adapted the leaf age algorithm from the Model of Emissions of Gases and Aerosols from Nature (MEGAN) with observed changes in Es across leaf ages. Our results showed that the variability in isoprene mixing ratios was higher between seasons (max during the dry-to-wet transition seasons) than between years, with values from the extreme 2015 El Niño year not significantly higher than in normal years. In addition, model runs considering in situ observations of canopy Es and the modification on the leaf age algorithm with leaf-level observations of Es presented considerable improvements in the simulated isoprene flux. This shows that MEGAN estimates of isoprene emission can be improved when biological processes are mechanistically incorporated into the model.