A global ocean general circulation model (OGCM) simulated interannual variability in the upper Indian Ocean between January 1982 and December 1997. The model is forced by NCEP wind stresses and the surface fluxes are provided by coupling the OGCM to an atmospheric boundary-layer model. The simulated interannual variability along two hydrographic lines in the Indian Ocean was compared to observations from expendable bathythermorgraph (XBT) data and satellite altimetry data. In both space-and timescales they were in good agreement. However, averaged correlation coefficients along the XBT lines for simulated and observed anomalies of sea-surface height (r&blmac;=0.52) are higher than those for anomalies of sea-surface temperature (r&blmac;=0.38). Possibly this is because the temperature anomalies are determined by interior ocean dynamics plus ocean-atmosphere heat fluxes (which have observational and parameterization errors), whereas the height anomalies represent a vertically integrated property of the ocean in which surface heat fluxes are less important. The results demonstrate that our model is currently better at representing the subsurface thermal structure than it is at modelling sea-surface temperature. It emphasizes the need for improved heat flux products. ¿ 1999 American Geophysical Union