Significant wave height and wind speed fields from ERA-40 are validated against buoy, ERS-1, and Topex altimeter measurements. To do so, we propose and apply a triple collocation statistical model. The model takes into account the random errors in observations and model results and allows the estimation of the variances of the errors. We first examine the case where the random errors of the different systems are independent, but situations where independence is not strictly observed are also considered. The results show that the ERA-40 predictions underestimate high values of significant wave height and, contrary to what would be obtained by less sophisticated statistical methods, wind speed, that the variance of the errors associated with the ERA-40 system is much higher than that of the errors of the measurements, and that the former shows a dependence on the value of the observations not present in the latter. The altimeter measurements of significant wave height are very precise, in contrast to the large uncertainty associated with the altimeter retrieved wind speeds.