thought to play a critical role in enzyme activation within the cell. In general, poor correlation exists between the serum magnesium concentration and intracellular levels. Lymphocyte magnesium does not correlate well with serum or red blood cell magnesium levels (Elin and Hosseini, 1985; Ryzen et al., 1986) and serum magnesium concentration does not accurately reflect muscle magnesium content (Alfrey et al., 1974; Wester and Dyckner, 1980). In normal subjects lymphocyte and skeletal muscle magnesium did correlate well, however not in patients with congestive heart failure (Dyckner and Wester, 1985). Thus, further evaluation is needed before lymphocyte or muscle magnesium content can be utilized to assess magnesium status.

Red blood cell magnesium has been determined by nuclear magnetic resonance (Rude et al., 1991). The fluorescence probe has been utilized for determination of free magnesium in lymphocytes and platelets (Hua et al., 1995; Nadler et al., 1992). Red blood cell magnesium values fall within days following institution of a low-magnesium diet in normal subjects (Fatemi et al., 1991). The mean intracellular free magnesium was lower than normal in patients at high risk for magnesium depletion (for example, individuals with diabetes or alcoholism) (Hua et al., 1995; Nadler et al., 1992; Rude, 1991).

In one study, total red blood cell magnesium concentration was found to be lower in elderly subjects (77.8 ± 2.1 years), as compared with younger people (36.1 ± 0.4 years), when the mean magnesium consumption of both groups was 311 ± 21 mg (13.0 ± 0.9 mmol)/day (Paolisso et al., 1992). However, the evidence was not judged sufficient to use red blood cell magnesium as an indicator of status.

The principal measure of adequate dietary magnesium in the past has been the dietary balance study (Greger and Baier, 1983; Hunt and Schofield, 1969; Mahalko et al., 1983; Schwartz et al., 1984, 1986). Most balance studies were performed at clinical research centers where the diet was constant and controlled. However, this technique still presents several problems, including the measurement of magnesium intake and urine and stool magnesium excretion. In addition, sweat and dermal losses of magnesium have not usually been considered.

In adults, balance studies should be performed at magnesium intakes just below and above those at which zero balance is achieved to obtain the approximately linear dependence of loss and reten-