We investigated the tsunami recorded at Monterey, California, during the 1989 Loma Prieta earthquake (MW=6.9). The first arrival of the tsunami was about 10 min after the origin time of the earthquake. Using an elastic half space, we computed vertical ground displacements for many different fault models for the Loma Prieta earthquake, and used them as the initial condition for computation of tsunamis in Monterey Bay. The synthetic tsunami computed for the uniform dislocation model determined from seismic data can explain the arrival time, polarity, and amplitude of the beginning of the tsunami. However, the period of the synthetic tsunami is too long compared with the observed. We tested other fault models with more localized slip distribution. None of the models could explain the observed period. The residual waveform, the observed minus the synthetic waveform, begins as a downward motion at about 18 min after the origin time of the earthquake, and could be interpreted as due to a secondary source near Moss Landing. If the large scale slumping near Moss Landing suggested by an eyewitness observation occurred about 9 min after the origin time of the earthquake, it could explain the residual waveform. To account for the amplitude of the observed tsunami, the volume of sediments involved in the slumping is approximately 0.013km3· Thus the most likely cause of the tsunami observed at Monterey is the combination of the vertical uplift of the sea floor due to the main faulting and a large scale slumping near Moss Landing.