We carried out a randomized controlled experiment in the Sikasso Region of Mali to test the hypotheses that (a) pre– and post-harvest training lowers the probability of crop losses and reduces household food insecurity; (b) post-training reminders have an even greater impact on reducing crop losses and lowering food insecurity; and (c) post-training reminders during specific times when such information needs to be applied reduced the probability of crop losses and lowers food insecurity over and above what could be achieved through training only. Aside post-harvest losses for which we could not reject the null hypothesis that post-training reminders have no impact, we easily reject the null that farmers who received only faceto- face training and those who received post-training reminders have identical outcomes. The null hypothesis that farmers who received only training are not different (with respect to our outcomes of interest) from those who received neither training nor reminders could not be rejected, except in the case of timely harvesting. These results suggest that going beyond training to providing timely reminders through voice messages could boost adoption and improve household welfare. Besides, the reminders intervention is relatively inexpensive, compared with traditional extension approaches.