Background and Aims: A 3D modelling approach simulating canopy structure was used in combination with a radiative transfer model to simulate light interception, distribution and microclimate in the fruiting zone. Methods and Results: This model was parameterised for four training systems (two vertical shoot-positioned systems with one or two pairs of catch wires (VSP-1W, VSP-2W); two non-shoot-positioned systems, gobelet (GOB) and bilateral free cordon (BFC)) and two cultivars (Syrah and Grenache). Light interception and canopy microclimate depended on the interactions between the intrinsic architecture of the cultivars and canopy manipulations. Shoot vigour and leaf area were the main determinants of canopy radiative balance. However, differences in shoot architecture accounted for up to 25% of the difference in light interception between cultivar × trellis system pairs at a given leaf area index (LAI). Light interception efficiencies and the proportion of sunlit leaf area (SLA) were 25–30% lower for VSP-2W than for BFC for intermediate LAI values. The genotypic differences in the ability to capture light were mostly induced by the 'procumbent' habit of the Syrah shoots. For this cultivar, shoot-positioned systems resulted in lower levels of fruit illumination at midday than the BFC and GOB systems, whereas the use of a catch wire in VSP-Grenache canopies made it possible to maintain light penetration in the fruit zone. Conclusions: These results highlight the problem of adapting the training system to both the architectural characteristics of the cultivar and climate. Free-standing systems had greater light interception and SLA than shoot-positioned systems. They may enhance fruit illumination for cultivars with 'procumbent' shoots. Significance of the Study: Non-positioned shoot systems offer the possibility of combining a high level of light interception, favourable microclimate and reduced labour-intensive practices for vineyards in conditions of moderate vigour.