We report on the use of the EGS4/BEAM Monte Carlo technique to predict the output factors for clinically relevant, irregularly shaped inserts as they intercept a linear accelerator's electron beams. The output factor for a particular combination--energy, cone, insert, and source-to-surface distance (SSD)--is defined in accordance with AAPM TG-25 as the product of cone correction factor and insert correction factor, evaluated at the depth of maximum dose. Since cone correction factors are easily obtained, we focus our investigation on the insert correction factors (ICFs). An analysis of the inserts used in routine clinical practice resulted in the identification of a set of seven "idealized" shapes characterized by specific parameters. The ICFs for these shapes were calculated using a Monte Carlo method (EGS4/BEAM) and measured for a subset of them using an ion chamber and well-established measurement methods. Analytical models were developed to predict the Monte Carlo-calculated ICF values for various electron energies, cone sizes, shapes, and SSDs. The goodness-of-fit between predicted and Monte Carlo-calculated ICF values was tested using the Kolmogorov-Smirnoff statistical test. Results show that Monte Carlo-calculated ICFs match the measured values within 2.0% for most of the shapes considered, except for few highly elongated fields, where deviations up to 4.0% were recorded. Predicted values based on analytical modeling agree with measured ICF values within 2% to 3% for all configurations. We conclude that the predicted ICF values based on modeling of Monte Carlo-calculated values could be introduced in clinical use.