The effects of infection by Mycobacterium avium paratuberculosis (MAP) on performance have to be estimated in order to assess its economic impact. This study aimed at quantifying the variation in fertility of dairy cows according to their MAP-infection status. The hypothesis of an indirect effect was set. Fertility was measured by the non-return rate at first and second services. A non-return was defined as the absence of another artificial insemination (AI) after the first one while the cow was still present. Three different statuses were defined based on both individual and herd results: positive cow, negative cow in a negative herd and negative cow in a positive herd. 27 612 AI from 72 135 cows in 1 470 herds were studied by logistic regression after adjusting on known factors influencing reproduction. Non-return rate was higher for infected cows compared to negative cows from negative herds (OR of 1.14, or +3.2 point of % of non-return rate). This increase was higher for parity 1 cows (OR of 1.20, or +4.4 point of % of non-return rate) compared to older cows. The effects were lower when comparing positive cows to negative cows in the same herds. Looking at these observations, the hypothesis of MAP-effect based on the relation between MAP-infection, production and reproduction is formulated. Due to the lack of protein absorption in the intestine, the milk production is reduced. In the subclinical stage of the disease, this could lead to a lower negative energy balance that could be associated with improved fertility.