This study presented an uncertainty assessment of the high-resolution global analysis of daily-mean ocean-surface vector winds (1987 onward) by the Objectively Analyzed air-sea Fluxes (OAFlux) project. The time series was synthesized from multiple satellite sensors using a variational approach to find a best fit to input data in a weighted least-squares cost function. The variational framework requires the a priori specification of the weights, or equivalently, the error covariances of input data, which are seldom known. Two key issues were investigated. The first issue examined the specification of the weights for the OAFlux synthesis. This was achieved by designing a set of weight-varying experiments and applying the criteria requiring that the chosen weights should make the best-fit of the cost function be optimal with regard to both input satellite observations and the independent wind time series measurements at 126 buoy locations. The weights thus determined represent an approximation to the error covariances, which inevitably contain a degree of uncertainty. Hence, the second issue addressed the sensitivity of the OAFlux synthesis to the uncertainty in the weight assignments. Weight perturbation experiments were conducted and ensemble statistics were used to estimate the sensitivity. The study showed that the leading sources of uncertainty for the weight selection are high winds (>15 ms?1) and heavy rain, which are the conditions that cause divergence in wind retrievals from different sensors. Future technical advancement made in wind retrieval algorithms would be key to further improvement of the multisensory synthesis in events of severe storms.