Model-based predictions were critical in eliciting a vigorous international public health response to the 2014 Ebola Virus Disease outbreak in West Africa. Here, we describe the performances of an extension of the CDC-initiated EbolaResponse Modeling tool to the Ebola Forecasting Challenge, which offered a controlled environment for epidemiological predictions. In the EbolaResponse tool, transmission risks and proportions of population affected by interventions were fitted to data via least square fitting. Prediction performances were evaluated for 5 prediction time points of 4 synthetic outbreaks. One-to-four week-ahead incidence predictions were well correlated with synthetic observations (rho ∼0.8), and overall ranking averaged over various error metrics was 4th of 8 teams participating in the context. EbolaResponse yielded moderately accurate predictions for final size, peak size and timing. The relative success of this easily adaptable mechanistic model, with reassessment of model parameters at fixed intervals, indicates that it can generate relatively accurate short-term forecasts, especially when interventions are staggered. An important downside of the model includes a lack of uncertainty estimates in its current framework. Overall, our results align with the conclusion that simple models with few parameters perform well for short-term prediction of epidemic trajectories.