We have designed a borehole tiltmeter using two horizontal pendulums which have periods of 1 s. We have installed the instruments at seven sites in Colorado and Wyoming to evaluate the secular tilt, the tides, and the coherence between nearby instruments. Using 28 days of data from Boulder, Colorado, the standard deviations of the tidal estimates are 5% for the semidiurnal component M2 and 9% for the diurnal component O1. The estimates agree with models that include the body tide, the ocean load, and the topographic correction to better than the estimated uncertainty. Tidal measurements at Erie, Colorado, have larger, possibly nonrandom variability that may be caused by a coupling between the tides and long-period tilts. The coherence between nearby instruments is >0.5 for frequencies ranging from 0.5 to 2 cycles per day, but drops to a much smaller value outside of this range. The secular tilt rate is quite variable and ranges from 0.1 to 1 μrad/yr. Measurements at Erie, Colorado, and in Yellowstone National Park, Wyoming, exhibit an annual or biannual periodicity with an amplitude of about to 1 μrad, which is consistent with a 1/f2 extrapolation of the diurnal noise power to longer periods. ¿ American Geophysical Union 1989