In several process plants the precision of parameter estimates is low because the installed set of instruments do not satisfy the new information requirements for different purposes, from simple data reconciliation to on-line optimization. This paper presents models to perform the upgrading of instrumentation at minimum cost to achieve maximum precision of selected parameters. Alternative equivalent models based on maximizing the precision of the parameters are capable on putting a bound on the capital cost. The intricacies of these apparently conflicting goals are explained and a unique procedure based on an MINLP model is presented.