Here we report on two field experiments from Lake Alpnach (surface area: 4.8 km2; maximum depth: 34 m) that were designed to study the process of boundary mixing and to estimate its efficiency, the ratio between the turbulent kinetic energy converted into potential energy and dissipated into heat, for diapycnal tracer transport. Lake Alpnach is particularly suited for this purpose because it is known from earlier experiments that (1) its currents follow a regular oscillatory pattern, associated with basin-wide standing internal waves (seiches, period ~0.5 days and 1 day, respectively), and (2) diapycnal tracer transport is mainly caused by boundary mixing. During one of the experiments reported here, strong seiches were excited regularly and damped on a timescale of the order of 3 days; during the other experiment seiching motion was comparably weak. If seiching is excited regularly, we find a persistent well-mixed bottom layer of 4--5 m height at the deepest part of the lake. In the absence of regular seiching the layer disappears within 10--20 days. On the sloping bottom boundaries the well-mixed layers (1) are of much more transient nature, (2) exhibit different thermal structure, and (3) decrease in thickness toward shallower depth inversely proportionally to the stability at the same depth in the lake interior. Their reduced thickness is possibly the result of repeatedly occurring intrusions of boundary mixed water masses that are observed to extend horizontally ~100--200 m into the lake interior. As a consequence of repeated generation of intrusions, mixing on the sloping boundaries is expected to be considerably more efficient compared to mixing over flat bottom boundaries. The observed mixing efficiency, the ratio of the rate of change of potential energy below depth z caused by turbulence to the energy loss by bottom friction below depth z, increases indeed from 0.01¿0.01 in the deepest well-mixed layers to 0.15¿0.04 in the upper pycnocline. ¿ 2000 American Geophysical Union