1. Neutral carrier-based liquid membrane micro-electrodes were constructed which are suitable for continuous measurements of [Mg(2+)](i) in cardiac and skeletal muscle preparations.2. The electrodes show a Nernstian behaviour in pure MgCl(2) solutions. In the presence of a constant ionic background chosen to simulate the cytoplasmic composition, the calibration function flattens progressively with lower [Mg(2+)], due to the interference of K(+) and Na(+). The response to changes in [Mg(2+)] is less than 0.5 sec.3. In quiescent preparations at room temperature (23 degrees C), the following basal [Mg(2+)](i) were determined: 3.5 mM (sheep Purkinje fibres), 3.1 mM (sheep ventricular muscle), 3.0 mM (ferret ventricular muscle) and 3.3 mM (frog skeletal muscle).4. In cardiac tissue, electrical stimulation does not measurably affect the basal [Mg(2+)](i).5. In the presence of 0.5 mM-Mg(2+) (o), the calculated Mg(2+) equilibrium potentials, E(Mg), are in the range of -23 to -25 mV, suggesting that Mg(2+) is not passively distributed across the sarcolemma in striated muscle.6. Further studies were performed on sheep Purkinje fibres to investigate the effect of various experimental interventions on [Mg(2+)](i).7. Elevating [Mg(2+)](o) from 0.5 to 10 mM resulted in a reversible increase of [Mg(2+)](i). The initial rate of increase corresponds to a Mg(2+) influx of 0.42 p-mole/cm(2).sec, or a magnesium permeability, P(Mg), of 1.6 x 10(-8)cm/sec.8. Increasing P(CO) (2) from nominally 0 to 100 mmHg (Tris-buffered vs. bicarbonate-buffered Tyrode solution) produced a reversible decrease in [Mg(2+)](i) by roughly 0.45 mM, probably due to Mg(2+) binding the newly formed intracellular HCO(3) (-).9. The effect of metabolic poisoning on [Mg(2+)](i) was assessed by exposure to cyanide, iodoacetic acid and 2-4-dinitrophenol. No significant increase in [Mg(2+)](i) indicative of a liberation of Mg(2+) from ATP was observed.