The magnitudes of bulk air-sea surface flux estimates calculated using three temporal averaging methods were compared. The reference method is a simple average of fluxes computed from hourly values of bulk meteorological parameters termed the sampling method (SM). In contrast, the scaler averaging method (SAM) computes the average flux from the average of the bulk data; thus it ignores correlations between variables. The vector averaging method (VAM) is similar to the SAM but uses the magnitude of the average wind vector rather than the average magnitude of the hourly wind vectors. The data used in this study were collected during the Tropical Ocean and Global Atmospheric pilot cruise of the R/V Wecoma in the equatorial Pacific Ocean near 0¿N, 145¿E from February 17 to March 10, 1990. The ratios of the SAM and VAM values relative to the SM values were studied as a function of averaging periods from 2--72 hours. The ratios vary little for times exceeding 36 hours. For averaging periods of 72 hours, the SAM estimates of QE, QH, &tgr;x, and &tgr;y were 102%, 61%, 21%, and 69% of the SM estimates, respectively; the VAM ratios were even lower. These results suggest that air-sea surface scaler fluxes and stress components computed from monthly, weekly, and even daily averaged bulk meteorological parameters can be seriously in error in equatorial, temporally variable wind regimes. ¿ American Geophysical Union 1993