We present our first results of the application of the measurement of atmospheric emission to correct for atmospheric phase variations in very long baseline interferometry. This technique corrects for the fluctuations in the phase path length through the atmosphere by measuring the water vapor content through its emissivity. Two epochs of 3-mm wavelength (λ) observations under different atmospheric conditions show that total power monitoring at Hat Creek and Kitt Peak Observatories can be used to reduce the rms interferometer phase when liquid water is not present in the beam. In the best cases, the root-mean-square (rms) phase is reduced by a factor of 2 to 1 rad. Instability in the receiver gain at the level of one part in 103 creates the 1-rad limit in the rms phase. The technique is currently most successful at eliminating variability on timescales greater than 10 s. However, it is not always successful, and we discuss sources for the residual variability in the phase. We also derive the expected value of the scale factor relating delay to brightness temperature at 86 and 230 GHz. As this technique is improved, future experiments will have greater sensitivity to weak sources, accurate astrometry and geodesy will be possible, and imaging at 1-mm λ may be explored.¿ 1997 American Geophysical Unio