Background The introduction of viral transneuronal tracers in the toolbox of neural tract-tracing methods has been an important addition in the field of connectomics for deciphering circuit-level architecture of the nervous system. One of the added values of viral compared to conventional retrograde tracers, in particular of rabies virus, is to provide a Golgi staining-like view of the infected neurons, revealing the thin dendritic arborizations and the spines that are major post-synaptic seats of neuronal connections. New method Here, we comparatively illustrate the characteristics of the labeling obtained in the same model system, the basal ganglia circuitry, by different retrograde viral tracing approaches, using the Bartha strain of pseudorabies virus, the SAD and CVS strains of rabies virus and by the conventional retrograde tracer cholera toxin B. To best contrast the differences in the capacity of these tracers to reveal the dendritic morphology in details, we focused on one population of first-order infected neurons in the striatum, which exhibit high spine density, after tracer injection in the substantia nigra. Results and conclusion None of the viruses tested allowed to detect as many neurons as with cholera toxin B, but the SAD and CVS strains of rabies virus had the advantage of enabling detailed Golgi-like visualisation of the dendritic trees, the best numerical detection being offered by the transneuronal rCVS-N2c-P-mCherry while poor labeling was provided by rCVS-N2c-M-GFP. Results also suggest that, besides different viral properties, technical issues about constructs and detection methods contribute to apparently different efficiencies among the viral approaches.