1. Potassium currents were measured in the extensor digitorum longus muscle of normal and mdx mice, which lack the protein dystrophin, using the cell-attached and inside-out patch clamp techniques, in the presence of asymmetrical K+ concentrations (3 mM in the pipette, 160 mM in the bath). 2. In cell-attached patches, the delayed rectifier was the most commonly found potassium channel, with a density of roughly 8 channels microns-2. Outward macroscopic currents were activated in macropatches depolarized to potentials positive to -60 mV. The probability of opening reached half-maximal values around -40 mV for control patches and -31 mV for patches from mdx mice. 3. Tail currents were linear in the range between -60 and +20 mV, reversing close to -100 mV. The single channel current at 0 mV, estimated from non-stationary analysis of variance, was used in conjunction with the slope of the linear part of the tail current to calculate the single channel conductance, yielding a value of 19 +/- 1 pS. 4. At 0 mV, the delayed rectifier inactivated with two time constants, of 70 +/- 20 ms and 600 +/- 200 ms. Prepulses of 500 ms duration to different potentials produced incomplete inactivation with inactivation reaching 50% of its maximum at -50 mV. 5. Single channel activity was recorded using small pipettes. Both single channel conductance and kinetic behaviour were in agreement with the macroscopic current data. 6. In excised patches, the delayed rectifier current ran down, unmasking other K+ channels. A Ca(2+)-dependent K+ channel of 186 pS (BK-like channel) was found frequently in patches bathed in solutions containing appropriate concentrations of calcium, especially at stronger depolarizations. A K+ channel of 63 pS was unmasked in control excised patches bathed in solutions devoid of ATP. This channel was not observed in patches excised from mdx fibers.