R/V Franklin followed a drifter for 8 days in the equatorial Indian Ocean to test the accuracy of mixed-layer heat and freshwater budget closure. Four-hourly triangles were repeated, towing a SeaSoar with a photometer and conductivity-temperature-depth (CTD) profiler. Currents relative to the buoy at one depth were obtained from (a) a current meter at 25 m below the buoy; (b) a Global Positioning System estimate; (c) a drag estimate. Acoustic Doppler current profiler (ADCP) shears provided currents at other depths. (b) and (c) agreed closely but differed from (a), possibly due to severe conditions at the current meter. Vertical advection was estimated from ADCP divergence around each triangle. Measurements on the buoy and the ship, calibrated earlier in the Coupled Ocean-Atmosphere Response Experiment (COARE), provided bulk surface fluxes. Accuracy of net surface heat fluxes was estimated at ¿10 W/m2. Rainfall differed considerably between the buoy and ship, but other quantities including shortwave radiation were very similar, even on cloudy days. A formalism for the budgets above a given isopycnal is developed, to clearly distinguish horizontal and vertical advection. When (b) and (c) above were used with the best estimate of surface fluxes, the 8-day heat budget misclose for water above the 21.5 isopycnal was (-4.6¿5) W/m2; the diurnal heating cycle was well resolved. The freshwater budget misclose was (0.9¿3.5) mm/d. A small amount of vertical diffusion, in the ratio determined by the slope of the T-S curve, will reduce both misclosures. Cruise design considerations for minimizing errors in the advection and storage components of heat and freshwater budgets are discussed. ¿ 1999 American Geophysical Union