Quantitative data on thalamocortical synapses in adult mouse barrels have been obtained largely by using lesion-nduced degeneration to label thalamic afferents. By the time degenerating axons can be identified with the electron microscope, they have broken up into many separate pieces, making it impossible to assess the distribution of synapses along unbroken lengths of afferent. Here, this deficiency is rectified by examining intact lengths of axon labeled by the injection of biotinylated dextran amine into ipsilateral thalamus. Serial thin section reconstructions were analyzed to determine the numbers of synapses per axon length made with dendritic spines vs. shafts and the locations of synapses with respect to axonal varicosities. Results for seven axonal segments from six mice showed an average of 0.2 synapses/microm; 80% were made with spines and 20% with dendritic shafts. Just over two-thirds of axonal varicosities formed one synapse; most of the remainder formed two and rarely three, whereas 8% formed none. Although most synapses occurred at varicosities (88%), more than 12% were made at cylindrically shaped regions of the reconstructed axonal segments. These results serve as a caveat for the use of light microscopy to quantify synapses, wherein the usual approach is to equate one varicosity with one synapse. For thalamocortical afferents to mouse barrels, equating one varicosity with one synapse would prove to be incorrect more than 30% of the time and would exclude the roughly 12% of synaptic connections made at cylindrical regions of thalamocortical afferents.