Monthly interannual anomalies of tropical Pacific sea level height from TOPEX/Poseidon altimetry are compared with simulation and assimilation products from a variety of models, ranging from a simple linear long wave approximation to ocean general circulation models. Major spatial similarities in the error patterns are identified. These include zonally elongated maxima in the northwest and southwest tropical Pacific Ocean, a band of high values near 10°N, slightly inclined toward the equator from the Central American coast, and low values on the equator and in the southeastern tropical Pacific. These features are also present in the pattern of small-scale variability (SSV) of sea level height. Spatial and temporal components of this SSV are analyzed for predominant variability types. Monte Carlo experiments identify the areas where high SSV is wind-driven, caused by a similar pattern of variability in the wind stress. Model products systematically underestimate signal variance in such areas. Variability in other areas is due to the instability of ocean currents. The major component of uncertainty in the gridded satellite altimeter analyses is due to sampling error, for which estimates are developed and verified.