In optimization in Rn with m nonlinear equality constraints, we study the local convergence of reduced quasi-newton methods, in which the updated matrix is of order n–m. Furthermore, we give necessary and sufficient conditions for superlinear convergence (in one step) and we introduce a device to globalize the local algorithm. It consists in determining a step along an arc in order to decrease an exact penalty function and we give conditions so that asymptotically the step-size will be equal to one. © 1989, Taylor & Francis Group, LLC.