Backstresses, associated with certain dislocation arrangements and their inter-dislocation long-range stresses, are known to contribute significantly to deformation response of metals, including kinematic hardening, the Bauschinger effect (BE) and the Hall-Petch effect. Various methods have been employed to quantify these backstresses at the macro-scale. One of these approaches, which has received relatively little attention, is the stress dip test. The strain rate observed during a load dip and hold, after previous plastic deformation, can be positive or negative, depending upon the level at which the load is held, and the relative magnitudes of competing friction and backstresses. The most direct interpretation of previously reported tests indicates a surprisingly high level of backstress in common materials, and which is generally also higher than the value extracted from an unload-reload test. In this paper, stress dip tests are performed on pure polycrystalline tantalum, along with unload-reload tests. A plateau is seen in the strain rate observed during the stress dip test, which has not been previously reported. If the backstress is interpreted to correspond with the stress level associated with the middle point of the plateau, in line with the friction/backstress model of the unload-reload test, the resulting backstress obtained from both tests are very similar. A novel crystal plasticity model, incorporating backstress, reversible dislocations and non-Schmid effects, is used to help justify this new approach. The model predicts the observed plateau in strain rate, and provides a slip-level interpretation of the macroscopically observed backstress. The slip-level backstress (when considered as a fraction of the stress prior to the dip) is reasonably similar to the values interpreted from the dip test experiment. The ∼23% lower value in the simulation may be due to the lack of certain aspects of the actual physics in the model.