A model that is too simple to capture the underlying model is likely to have high bias and low variance (underfitting). Overly complex models typically have low bias and high variance (overfitting).