Marine boundary layer ozone seasonal cycles have been quantified by fitting the sum of two sine-curves through monthly detrended observations taken at three stations: Mace Head, Ireland and Trinidad Head, California in the northern hemisphere and Cape Grim, Tasmania in the southern hemisphere. The parameters defining the sine-curve fits at these stations have been compared with those from a global Lagrangian chemistry-transport model (STOCHEM-CRI) and from fourteen ACCMIP chemistry-climate models. Most models substantially overestimated the long-term average ozone levels at Trinidad Head whilst they performed much better for Mace Head and Cape Grim. This led to an underestimation of the observed (North Atlantic inflow – North Pacific inflow) difference. The models generally under-predicted the magnitude of the fundamental term of the fitted seasonal cycle, most strongly at Cape Grim. The models more accurately reproduced the observed second harmonic terms compared to the fundamental terms at all stations. Significant correlations have been identified between the errors in the different models’ estimates of the seasonal cycle parameters; these correlations may yield further insights into the causes of the model – measurement discrepancies.