Static estimators are commonly used as "soft sensors" in the process industry. The performance of the estimators is dependent on whether it is used for monitoring (open-loop) or for closed-loop control applications. In this work, we propose to design estimators that are specialized for each case. The approach is to minimize the estimation error for expected disturbances and measurement noise. The main extension, compared to previous work, is to include measurement. noise and to provide explicit formulas for computing the,optimal static estimator. We also compare the results with standard existing estimators (e g, partial least squares (PLS)). The approach is applied to estimation of product composition in a distillation column from a combination of temperature measurements.