A mathematical model for a boiling water reactor steam-turbine cycle was assembled by means of a configurable, steady-state modeling tool TEMPO. The model was connected to live plant data and intermittently fitted to these by minimization of a weighted least-squares object function. The improvement in precision achieved by this reconciliation was assessed from quantities calculated from the model equations linearized around the minimum and from Monte Carlo simulations. It was found that the inclusion of the flow-passing characteristics of the turbines in the model equations significantly improved the precision as compared to simple mass and energy balances, whereas heat transfer calculations in feedwater heaters did not. Under the assumption of linear model equations, the quality of the fit can also be expressed as a goodness-of-fit Q. Typical values for Q were in the order of 0.9. For a validated model Q may be used as a fault detection indicator, and Q dropped to very low values in known cases of disagreement between the model and the plant state. The sensitivity of Q toward measurement faults is discussed in relation to redundancy. The results of the linearized theory and Monte Carlo simulations differed somewhat, and if a more accurate analysis is required, this is better based on the latter. In practical application of the presently employed techniques, however, assessment of uncertainties in raw data is an important prerequisite.