Although widely used elsewhere, multivariate analysis of temperature is not often explored to evaluate temperature patterns in Patagonia, the southern tip of South America. Since there is no general agreement in locating this region within a global temperature-based climatic scheme, classifying temperature regimes using cluster analysis can lead to a better understanding of the Patagonian climate. Working with 14-year long series of daily temperature data in 21 weather-stations south of 40Β°S, an average linkage clustering is presented within the matrix of correlation among the departures from the monthly temperature averages. Four clusters were identified at the r =0.55 level, the mapping of which is consistent with some ecoclimatic boundaries. This was an unexpected result as deviations rather than actual temperature levels were employed to perform the clustering. Also, the two major clusters adopted a zonal disposition regardless of the main climatic division in Patagonia, i.e. west -east. In this way it is shown that this orography-induced boundary, is merely rain-dependent with little consequence to the temperature regime. Standard deviation of mean annual temperature was also mapped as a complementary tool to depict the oceanic -continental continuum in the area, upon which no uniform criterion exists. Finally, the weather-station that best reflects the interannual variation of temperature was identified in each one of the two main clusters. The correlation among both base stations and the others, showed two extensive homogeneous areas that cover most of Patagonia, while two marginal areas exhibit separate patterns.