By measuring the ''scaled dissipation ratio'' &Ggr;, which is the relative magnitude of the dissipation of thermal variance compared with the dissipation of turbulent kinetic energy, one can distinguish between salt fingering and nonfingering regimes. This is illustrated by comparing data from three test cases to predictions from (1) a mixing model which considers turbulence only and (2) a model which describes salt fingers as the sole mixing process. In a turbulent, nondouble-diffusive surface layer we find that measurements of &Ggr; are very close to the value predicted by the turbulent mixing model. A contrasting case is provided by a themohaline staircase located at 1200- to 1330-m depth. There the salt finger model provides a better description of the mixing than does the turbulent mixing model. The third case study is of measurements between 150 and 400 m where the water column is characterized by low density ratios and ''intermittent steppiness' in temperature and salinity profiles. Here, values of &Ggr; are inconsistent with a model containing only salt finger or turbulent mixing; instead, the observations suggest that both processes are important. Using the observed values of &Ggr; in a combined model suggests that 24% of the observed turbulent kinetic energy dissipation is due to salt fingers. A corresponding estimate of the vertical eddy diffusivity of salt (and nutrients) is 2 times larger than that computed from the turbulence-only mixing model and 50% larger than the vertical eddy diffusivity for heat as determined by the Osborn-Cox relation. ¿ American Geophysical Union 1993