The recent increase in food prices has revealed that climate, combined with an expanding population and a widespread change in diet, may result in an end to an era of predictable abundance of global cereal crops. The objective of this paper is to estimate changes of agriculturally-relevant growing season parameters, in-cluding the start of the season, length of the growing period and the position of the height or peak of the sea-son, in the primary regions with rainfed agriculture during the past 26 years. Our analysis found that globally, 27% of cereal crop areas have experienced changes in the length of the growing season since 1981, the ma-jority of which had seasons that were at least 2.3 days per year longer on average. We also found both neg-ative and positive trends in the start of season globally, with different effects of changing temperature and humidity being isolated depending on the country and region. We investigated the correlation between the peak timing of the growing season and agricultural production statistics for rain fed agriculture. We found that two thirds of the countries investigated had at least 25% of pixels with crop production that be-haved differently than expected from the null hypothesis of no correlation. The results show that variations in the peak of the growing season have a strong effect on global food production in these countries. We show that northern hemisphere countries and states appear to have improved model fit when using phenological models based on humidity while southern hemisphere countries and states have improved model fit by phe-nological models based on accumulated growing degree days, showing the impact of climate variability dur-ing the past two and a half decades. Published by Elsevier Inc.