Developing effective vaccines to protect the mammary gland (MG) against infections has remained an elusive goal. Tackling the issue of the elicitation of protective immunity in the MG with a new angle could help break the deadlock. The limited success of the past and current vaccine attempts could be a consequence of the modulation of the immune response (IR) in the MG, supposing that the MG has co-evolved towards a modus vivendi with MG-adapted bacteria that favors chronic subclinical infections. In this line of thought the tissue-specific IR is constrained in a way which would reduce the collateral damages of the inflammatory reaction, but would also blunt the efficiency of antibacterial effectors. This MG-specific orientation of the IR would also prevent the development of a strong adaptive immune response to invading bacteria, as shown by the recurrence of infection in a given MG with the same bacterial strain with similar course and outcome. Recently, we used a local (intramammary) immunization regime which allowed better control of an Escherichia coli infection while reducing the inflammation and clinical signs. Yet the protective response was associated with an increase in IFN-γ and IL-1β in milk compared to control cows. This indicates that improvement of the response efficiency is compatible with a controlled inflammatory reaction. Ideally, a good mastitis vaccine should elicit a balanced IR in order to control and clear the infection without causing scarring damages to the MG. To this end, we need to gain new knowledge of the MG antigen-presenting cells, the balance between suppressive and effector T cells, and of the local conditions that shape the IR. This would make it possible to overcome the local constraints of the IR by using the appropriate vaccination regime and adjuvant.