State-space models are used to examine long-term spatial patterns in a set of temperature time series representing the meridional and offshore extent of the California Current System (CCS). We use global one-degree summaries from the World Ocean Database at 11 locations and 10 standard depths in the upper 200 m for the period 1950--1993. Four common trends account for most of the total variance and the important time-dependent features of the temperature series. These trends also reveal that different geographical regions and depth strata within the CCS have clearly distinct temporal patterns of interannual to decadal thermal variability: the first common trend reveals a series-long warming tendency at all locations, with the greatest changes occurring in the upper 50 m (75 m) for the coastal (offshore) stations; the second common trend delineates a cross-shore separation, with the coastal locations having greater variance and a stronger impact from most El Ni¿o events and decadal-scale warming and cooling tendencies; the third common trend, with signals strongest in the thermocline, provides a measure of the northern extent of influence of El Ni¿o events and reveals greater interannual variance south of 38¿N; the fourth common trend separates the series by depth, revealing significant long-term trends in thermal stratification. These spatially-variant thermal fluctuations identify three characteristic patterns of El Ni¿o influence in the CCS, reflecting both the meridional extent and depth structure of their signals. La Ni¿a signals are less spatially variable in the CCS. We also describe the spatially heterogeneous anomaly structure associated with large-scale decadal climate shifts, such as occurred around 1976. It is shown that this event was a coastal- and surface-intensified acceleration of a warming trend that had begun several years earlier. We briefly discuss the biological consequences of the spatially and temporally varying thermal structure of the CCS revealed by this analysis.