In 1978, a 20 × 106-m3 landslide of ≈900-yr return period was reactivated near Waipara River, South Island, New Zealand, in lower Tertiary marine muds, marls, sands, and limestone. A central block moved 155 m, reaching ≈5 µm/s after ≈3 months, before slowing exponentially over 4 yr. Movement began after a wet winter of ≈100-yr return period, but a previous drought may have hastened the movement. The sequence of dry summer and wet winter had a 1 in 1500-yr recurrence interval. A favored hypothesis for the role of drought is desiccation cracking, allowing greater or more thorough wetting of impermeable regolith. Acid leaching of calcite as pyrite weathered and dilation of overcompacted mudstone also occurred. Movement began as the water store peaked at the end of winter and rising shear stress exceeded diminishing strength. A kinematic model showing viscosity, friction, and a linearly varying thrust explains most of the movement. Small spurts in movement, perhaps from growth of the head region, suggest that ultimate displacement was unpredictable. The net driving force (≈+10−2 to ≈−10−3 N) was so very small relative to gravity's downslope pull (≈1010 N) that a deterministic model including slope and gravity would require impossible precision (1:1012) to give useful predictions. If the preconditioning role of drought is not refuted, the antithesis of current palliatives (spray irrigation to prevent desiccation) may be needed to reduce the likelihood of recurrence.