The paper discusses short- and long-term probability models of ocean waves. The Gaussian theory is reviewed, and nonlinear short-term probability distributions are derived from a narrow band second-order model. The nonlinearity has different impact on different measurement techniques, and this is further demonstrated for wave data from the WAVEMOD Crete measurement campaign and laser data from the North Sea. Finally, we give some examples on how the short-term statistics may be used to estimate the probability distributions for the maximum waves during individual storms as well as in a wave climate described by long-term distributions. (C) 2000 Elsevier Science B.V. All rights reserved.