The paper deals with the problem of identifying stochastic unobserved twocomponent models, as in seasonal adjustment or trend-cycle decompositions. Solutions based on the properties of the component estimation errors are con sidered, and analytical expressions for the variances and covariances of the dif ferent types of estimation errors (errors in the final, preliminary, and concurrent estimator and in the forecast) are obtained for any admissible decomposition. These expressions are relatively simple and straightforwardly derived from the A r im a model for the observed series. It is shown that, in all cases, the estimation error variance is minimized at a canonical decomposition (i.e., at a decomposition with one of the components noninvertible), and a procedure to determine that decomposition is presented. On occasion, however, the most precise final estimator is obtained at a canoni cal decomposition different from the one that yields the most precise concurrent estimator. Three examples illustrate the results and the computational algorithms. The first and second examples are based on the so-called Structural Time Series Model and A r im a Model Based approaches, respectively. The third example is a class of models often encountered in actual time series.