The quantitative characterization of ion channel properties in pancreatic beta-cells under typical patch clamp conditions can be questioned because of the unreconciled differences in experimental conditions and observed behavior between microelectrode recordings of membrane potential in intact islets of Langerhans and patch recordings of single cells. Complex bursting is reliably observed in islets but not in isolated cells under patch clamp conditions. E. Rojas et al. (J. Membrane Biol. 143:65-77, 1995) have attempted to circumvent these incompatibilities by measuring currents in beta-cells in intact islets by voltage-clamping with intracellular microelectrodes (150-250 M omega tip resistance). The major potential pitfall is that beta-cells within the islet are electrically coupled, and contaminating coupling currents must be subtracted from current measurements, just as linear leak currents are typically subtracted. To characterize the conditions under which such coupling current subtraction is valid, we have conducted a computational study of a model islet. Assuming that the impaled cell is well clamped, we calculate the native and coupling components of the observed current. Our simulations illustrate that coupling can be reliably subtracted when neighbor cells' potentials are constant or vary only slowly (e.g., during their silent phases) but not when they vary rapidly (e.g., during their active phases). We also show how to estimate coupling conductances in the intact islet from measurements of coupling currents.