Models examining the present-day relationship between macro-scale patterns in terrestrial species richness and variables of water and energy demonstrate that a combined water–energy model is a good predictor of richness in mid-to-high latitude regions. However, the power of the individual water and energy variables to explain this richness through time has never been explored. Here, we assess how well energy and water can predict long-term variations in plant richness using a 320 000-year fossil pollen data set from Hungary. Results demonstrate that a combined water–energy model best explains the variation in plant diversity through time. However, this long temporal record also demonstrates that amplitude of energy variation appears to be a strong determinant of richness. Decreased richness correlates with increased climate variability and certain species appear to be more susceptible according to their ecological traits. These findings have important implications for predicting richness at times of increasing climate variability.