The Kupaianaha vent, the source of the 48th episode of the 1983-to-present Pu'u 'O'o-Kupaianaha eruption, erupted nearly continuously from July 1986 until February 1992. This investigation documents the geophysical and geologic monitoring of the final 10 months of activity at the Kupaianaha vent. Detailed very low frequency (VLF) electromagnetic profiles across the single lava tube transporting lava from the vent were used to determine the cross-sectional area of the molten lava within the tube. Combined with measurements of lava velocity, these data provide an estimate of the lava output of Kupaianaha. In addition, lava temperatures (calculated from analysis of quenched glass) and bulk-rock chemistry were obtained for samples taken from the tube at the same site. The combined data set shows the lava flux from Kupaianaha vent declining linearly from 250000 m3/day in April 1991 to 54000 m3/day by November 1991. During that time surface breakouts of lava from weak points along the tube occurred progressively closer to the vent, consistent with declining efficiency in lava transport. There were no significant changes in lava temperature or in bulk MgO content during this period. Another eruptive episode (the 49th) began uprift of Kupaianaha on 8 November 1991 and erupted lava concurrently with Kupaianaha for 18 days. Lava flux from Kupaianaha decreased in response to this new episode, but the response was delayed by approximately 1 day. After 14 November 1991, lava velocities were no longer measurable in the tube because the lava stream beneath the skylight had crusted over; however, the VLF-derived electrical conductances documented the decreasing flux of molten lava through the tube. Kupaianaha remained active, but output continued to decrease until early February 1992 when the last active surface flows were seen. In November 1991 we used the linearly decreasing effusion rate to accurately predict the date for the death of the Kupaianaha vent. The linear nature of the decline in lava tube conductance and the delayed and slow response of the Waha'ula tube conductances to the 49th eruptive episode led us to speculate that (a) the Kupaianaha vent shut down because of a decrease in driving pressure and not because of a freeze-up of the vent, and (b) that Pu'u 'O'o, episode 49, and Kupaianaha were fed nearly vertically from a source deep within the rift zone.