A series of observing system simulation experiments (OSSE) is performed on a simulated data set which was designed to mimic the wind-forced response of the tropical Pacific Ocean. This data set was constructed by adding random perturbations to the FSU monthly mean pseudostress anomaly data. These perturbed pseudostress anomaly fields were used to drive a simple linear model whose output was sampled to form synthetic observations. The statistics of the perturbations are given by estimates of the errors in the pseudostress data calculated in an earlier study. OSSE are performed using simulated sea level height data from island tide gauge stations and from selected TOGA-TAO (Tropical Ocean and Global Atmosphere Thermal Array for the Ocean) mooring sites. Data from the TOGA-TAO moorings are assumed, in one experiment, to consist of sea level height data, identical to that from tide gauges. In a further experiment, mooring data consist of amplitudes of the first two vertical modes. Errors in the OSSE are seen to be comparable to errors obtained in comparison to real data where such comparisons are available. Assimilation of data at six island stations results in noticeable, but not dramatic improvement in the analysis, as was noted in an earlier study with actual observations. OSSE using simulated mooring data showed accuracy in the sea level height field comparable to that of the instruments across much of the basin. Inclusion of the separate baroclinic modes resulted in negligible improvement. Simulated fields of 20¿ isotherm depth anomaly were also produced. Results were similar to the sea level height results. As in the simulation of sea level height, inclusion of the separate baroclinic modes resulted in negligible improvement. ¿ American Geophysical Union 1990