The reliability of relative vertical motions detected by levelings of bench marks installed at Pinyon Flat is highly dependent on the kinds of bench marks installed for monitoring purposes. The most stable marks are the end monuments for two 535-m-lonog fluid tilt-meters. Nine direct measurements of average slope between tiltmeter end monuments agree to withnn a standard deviation about the mean of 0.2 mm (0.4 μrads), comparable to the slope measured by the tiltmeters themselves. Motions of class B steel rod marks are random with standard deviations ranging from 0.5 to 3.2 mm, averaging 1.0 mm. The motions of nearly all rod marks decreased greatly in the fourth year, evidently because of a time-dependent establishment of stability in the decomposed granite fundament. Other kinds of bench marks are stable between 0.5 and 0.8 mm. Tectonic tilt is not revealed in 15 levelings of the array in 4 years, consistent with the tiltmeters, but the random bench mark motions dominate the errors in the estimated tilt.