We investigate the seasonal sea surface height (SSH) variability on large spatial scales in the North Atlantic by using both a numerical simulation and in situ data. First, an ocean general circulation model is run with daily forcing from the European Center for Medium-Range Weather Forecasts reanalysis. We evaluate the different contributions to the seasonal SSH variability resulting from the surface heat fluxes, advection, salt content variability, deep ocean steric changes, and bottom pressure variability. These terms are compared with estimates from in situ data. North of 20°N, there is an approximate balance between hQ, the air-sea heat flux induced changes in steric height, and SSH variability. The next important component is the advection (its contribution to the annual amplitude is of the order of 1 cm except near the western boundary); other contributions are found to be smaller. Between 10°N and 10°S the advection variability induced by the seasonal wind stress cycle is the primary source of SSH variability. We then compare the sea surface height annual harmonic from TOPEX/Poseidon altimetry with the steric effect from the heat flux and with model and/or in situ estimates of the other terms. In many areas north of 20°N the balance between hQ and the altimetric SSH seasonal cycle is closed within the uncertainty limit of each of the terms of the SSH budget. However, hQ and the SSH do not balance each other in the eastern North Atlantic, and the results are sensitive to the choice of the heat flux product, suggesting that significant errors, typically 20-40 W m^-2 for the seasonal cycle amplitude, are present in the meteorological model heat fluxes.