The design of metering plans for power system state estimation is a classical, complex, combinatorial optimization problem, many times solved by a family of metaheuristic algorithms. This paper presents a methodology for designing reliable and robust metering systems, taking into account aspects, such as: observability; absence of critical data; possible changes in network configuration; loss of remote terminal units. To accomplish this task, a simple, practical value, constructive heuristic that considers inherent characteristics of the meter placement problem is proposed. Simulations with the IEEE 14, 30, and 118-bus test systems, as well as with part of a real Brazilian system, illustrate the performance of the proposed constructive approach. Regarding redundancy requisites, high quality metering plans are obtained without requiring a formal optimization technique. It is also shown that the proposed heuristic can be employed in conjunction with an optimization tool (a popular metaheuristic, e.g. ant colony optimization) to achieve even better results. Test results obtained in the paper are encouraging, as compared with those found in the technical literature.