By considering the absorbed power and energy based on heating power profile during extraction, a generalized model with washing coefficient (b), diffusion coefficient (k) and predictive parameter namely absorbed power density (APD) was developed for microwave-assisted extraction (MAE) at any operational heating modes. To study the model, MAE of flavonoids from cocoa (Theobroma cacao L.) leaves was conducted using the heating modes (constant-power, two-steps-power, intermittent-power and constant-temperature) at various microwave power (100-300W) and extraction temperature (50 degrees C and 70 degrees C). The results shows that the model (b = 0.5595 and k' = 0.01279 mL/J) is able to predict the normalized extraction yields of MAE at any heating modes, heating power, microwave system, extraction scale and batch of plant sample with less than 4% discrepancy. The accuracy of the prediction relies on particle size of sample (0.25-0.60 mm), type of extraction solvent (85% aqueous ethanol) and solvent to feed ratio (50 mL/g). (C) 2015 Elsevier B.V. All rights reserved.