A statistical analysis of daily water-level changes in an abandoned coal mine indicates that precipitation affects the potentiometric level of the mine, independent of associated atmospheric pressure changes and changes in the water level of an overlying aquifer. The independent statistical effect of precipitation (0.99 cm of water-level change per centimeter of rainfall) is interpreted to reflect either lateral percolation from the coalbed's subcrop (1.2 km from the mine) or rapid recharge through mine-associated pathways, such as poorly plugged shafts, boreholes, or subsidence fractures. The relationship between water-level changes in the mine's voids and changes in the overlying aquifer is also statistically significant, but the regression coefficient (0.04) is an order of magnitude smaller than that for precipitation, indicating that vertical percolation (which is represented by covariance of the two aquifers) through undisturbed overburden may be less effective than the recharge associated with precipitation that bypasses the overburden. An equivalent analysis of water-level changes in an underlying unmined coalbed indicated that precipitation had a weaker direct effect (regression coefficient of 0.34, compared with 0.99), although it was still the dominant independent variable. In contrast, the effect of water-level changes in an overlying aquifer (the flooded mine itself) was relatively stronger (regression coefficient of 0.15, compared with 0.04), indicating that vertical percolation through interburden is more important at depth.