Presented at the 2014 AGU Fall Meeting702 CTD profiles were collected in the subtropical northern South China Sea at and in the vicinity of the SouthEast Asian Time-series Study (SEATS) station (18.2°N, 115.8°E) between 17.5 and 18.5°N and 115.3 and 116.3°E in 64 cruises in 1997 to 2013. The hydrographic structure of the upper water above the permanent thermocline may be classified into 4 principal types: (a) classic type (an almost isopycnic upper water); (b) stepwise type (with one or more small but significant step-increases in σθ in the upper water); (c) graded type (an approximately constant depth gradient in a monotonic increase in σθ in the upper water); and (d) mixed type (a combination of the stepwise and graded types). The 4 types of upper water were found in 75, 14, 5, and 6% of the cruises, respectively. Ten schemes were applied to these data to determine the mixed layer depth (MLD): 4 fixed temperature difference (FTD) methods (0.2, 0.5, 0.8 and 1.0°C decrease from 10 m); 1 fixed density difference (FDD) method (0.125 σθ increase from 10 m); 1 fixed temperature gradient (FTG) method (at 0.05°C/m); 3 fixed density gradient (FDG) methods (at 0.01, 0.05 and 0.1 σθ/m); and the maximum density gradient (MDG) method. MLD could not be clearly depicted in the 3 minor types of upper water. In the classical type, while similar MLD-s were found in a large majority of the cruises among all 10 methods, substantial discrepancies among methods could be found. The most consistent results, generally within ±5 m, were found among the FDG method at 0.05, 0.1 σθ/m and FTD method at 0.8 and 1.0°C. The MDG method gave consistently deeper MLD by ~8 m. If that difference was taken into account, the results were generally consistent with those from the other 4 methods. The remaining 5 methods could all yield MLD-s shallower than the first 4 methods by >10 m as they failed to capture the bottom of the mixed layer as indicated by visual inspection.