Agricultural commodity price spikes are very damaging to economies which depend on food imports to satisfy their food security needs but are difficult to properly anticipate by market fundamentals alone. In this chapter we quantify the occurrence of such events, and search for determinants or early warning signals. We show that the distribution of price returns are non-normal, implying that the extreme events are more frequent than what is implied by a Gaussian distribution of returns. We empirically show that while some macro variables, such as a lax monetary policy and previous commodity market volatility, may signal a possible extreme event such as a price spike, the predictability of such events is very difficult. We also show that a low realized volatility in a previous period seems to signal a high probability of a subsequent extreme price event. The results suggest that the non-normal nature of internationally traded food commodity products implies signihcant challenges for insuring food imports by food-insecure countries.