Central to the "weather-climate link" is the link between variations of extratropical weather statistics ("stormtracks") and global sea surface temperature (SST) changes on interannual and longer scales. In this study, the SST-forced stormtrack signal in each northern winter in 1950-2003 was estimated as the ensemble-mean stormtrack anomaly in atmospheric general circulation model (AGCM) integrations for that winter with prescribed observed SSTs. Since the stormtrack signals could not be derived directly from the archived monthly-mean AGCM output, they were diagnosed from the SST-forced winter-mean 200 mb height signals using an empirical linear stormtrack model (STM). For four particular winters, the El Nino of JFM 1987, the El Nino of JFM 1998, the La Nina of JFM 1989, and the La Nina of JFM 1999, the stormtrack signals were estimated directly, and more accurately, from additional large ensembles of AGCM integrations with archived daily output. The linear STM was remarkably successful at capturing the AGCM's stormtrack signal in these four winters, and was therefore also deemed suitable for estimating the signal in other winters. We conclude that a potentially predictable SST-forced stormtrack signal exists in many winters, but its strength and pattern can change substantially from winter to winter. The pattern correlation of the SST-forced and observed stormtrack anomalies is high enough in the Pacific-North American sector to be of practical use. In the Euro-Atlantic region, we find much lower correlations, which we attribute to substantial AGCM error in representing the Euro-Atlantic response to tropical SST forcing, rather than to intrinsically low stormtrack predictability.