Comparison of tree-ring-based warm-season temperature reconstructions and their instrumental target data reveals substantial divergence between (warmer) early instrumental measurements and (colder) proxy estimates. Here we detail this systematic misfit for the Northern Hemisphere before ~1900 and the European Alps before ~1850. Five hypotheses related to both proxy and target uncertainties are presented towards explaining this phenomenon. These include: (1) tree-ring detrending methods, (2) biological persistence in the proxy time-series, (3) uncertainties and instabilities in the growth response to given climatic parameters, (4) reduced instrumental station availability back in time, and (5) instrumental data homogeneity. We suggest that uncertainties in the choice of instrumental targets at the hemispheric scale, and instrumental data inhomogeneities at the Alpine and possibly also the hemispheric-scale are the most important factors in explaining this offset. Assessment of homogeneity at larger scales remains challenging. Attention is drawn to possible warm biases in early thermometer shelters and the relevance of proxy/target discrepancies for understanding and quantifying the amplitude of both recent anthropogenic and past natural forced climate fluctuations.