1: \begin{abstract}

2:   The Nonlinear Noisy Leaky Integrate and Fire (NNLIF) model is widely

3:   used to describe the dynamics of neural networks after a diffusive

4:   approximation of the mean-field limit of a stochastic differential

5:   equation system.  When the total activity of the network has an

6:   instantaneous effect on the network, in the average-excitatory case,

7:   a blow-up phenomenon occurs.  This article is devoted to the

8:   theoretical study of the NNLIF model in the case where a delay in

9:   the effect of the total activity on the neurons is added.  We first

10:   prove global-in-time existence and uniqueness of classical

11:   solutions, independently of the sign of the connectivity parameter,

12:   that is, for both cases: excitatory and inhibitory. Secondly, we

13:   prove some qualitative properties of solutions: asymptotic

14:   convergence to the stationary state for weak interconnections and a

15:   non-existence result for periodic solutions if the connectivity

16:   parameter is large enough.  The proofs are mainly based on an

17:   appropriate change of variables to rewrite the NNLIF equation as a

18:   Stefan-like free boundary problem, constructions of universal

19:   super-solutions, the entropy dissipation method and Poinca\-r\'e's

20:   inequality.

21: \end{abstract}