We develop a predictive model which has the ability to explain a postulated style of episodic plate tectonics on Venus, through the periodic occurrence of lithospheric subduction events. Present-day incipient subduction zones are associated with the existence of arcuate trenches on the Venusian lithosphere. These trenches resemble terrestrial subduction zones, and occur at the rim of coronae, uplift features thought to be due to deep-mantle convective plumes. The model we adopt represents the lithosphere as the thermal boundary layer which lies above a convective plume. We assume a temperature-dependent nonlinear viscoelastic rheology, and we assume a stress-based criterion for plastic yield. In developing this latter criterion, we are led to a re-interpretation of the strength envelope which is commonly used in analysing lithospheric stress, and we propose that the plastic yield strength has meaning (and is finite) below the lithosphere, using behaviour in the Earth as our 'laboratory' justification for this view. An inferred yield stress on the Earth is ca. 300 bar (30 MPa). Our model then shows that a thickening lithosphere becomes progressively more fluid as the stresses induced by the buoyant convective plume become large. Failure occurs when the effective lithosphere viscosity becomes equal to that of the underlying mantle. We show that reasonable expected values of yield stress in the range 100-200 bar (10-20 MPa) for Venusian mantle rocks are consistent within the framework of the model with radii of coronal trenches in the range 100-1200 km, and with the approximate time (200-800 Myr) which they may take to develop.