To better assess the occurrence of climate variability and change and related effects on crop production for improved adaptation in Benin, we analysed the stochastic dependence of dry and wet dekads (i.e. ten days), for two periods (1951-70 compared to 1971-1990) at 36 meteorological stations, based on Markov chains. The implications on maize (Zea mays L.) production, the main staple food in the country, was assessed by comparing drought probabilities with crop stage, water requirement and water stress. We found (i) a slight increase in the marginal probability of dry spells P(D) during the 1971-1990 period compared to the 1951-70 one, (ii) that the second period 1971-1990 was characterised by a general increase in the succession of dry states, a higher transition probability from a dry dekad to the next dry P(D|D) than all other transition probabilities and a significantly higher probability of transition from two subsequent dry dekads to a third dekad dry P(D|DD) than P (D|D) at most stations, (iii) an increase in the length of dry spells and, therefore, dry periods on the sub-period 1971-1990, and (iv) that maize crop during its development in Benin is more likely to be subject to dry dekads successions than wet ones, resulting in poor yield, especially after 1970. This study shows that the frequent maize yield decline in Benin is partly explained by an increased occurrence of the succession of dry dekads (i.e. increase in the length of dry spells).