In this paper, we study the problem of estimating a regression function in a convolution model. We consider the following model: $y(x_k)=h(x_k)+\varepsilon_k , h(x)=f\star g(x)=\int_{\R} f(x-y)g(y)dy, k=-n, \dots, n-1$ where $g$ is assumed to be known and $f$ is the unknown function to be estimated; the errors $(\varepsilon_k)_{-n\le k\le n-1}$ are independent and identically distributed (i.i.d.) such that $\mathbb{E}[\varepsilon_k]=0$ and $\mathrm{Var}(\varepsilon_k)=\sigma_\varepsilon^2<+\infty$, known; the points $(x_k=kT/n)_{-n\le k\le n-1}$ are deterministic and equispaced on the interval $[-T,T]$, where $0