In this paper we address the issue of the efficient estimation of the cointegrating vector in linear regression models with variables that follow general (higher order and fractionally) integrated processes.