Recently, a hybrid method has been developed, wherein the positions of some polymer segments are constrained to a small volume. The volume's position is sampled with a Monte Carlo algorithm. The distribution of the remaining segments is determined with Scheutjens–Fleer self‐consistent field theory (SF‐SCF). This incorporates thermal fluctuations into the model. Here it is investigated whether this also leads to an improved treatment of the excluded volume interactions. Dendrimers, with f = 1⋅⋅⋅7 generations, g = 2⋅⋅⋅5 spacers per branch point, and spacers of 20 or 50 segments, are used as a model system. The focus is on the radius of gyration, the asphericity, the radial density, and the end point and first branch point distribution. As expected, both SCF methods underestimate the radius of gyration due to underestimating the short range excluded volume. The SF‐SCF model, however, also gives a slightly different scaling and, in contrast to the other models, also predicts a bimodal distribution for first generation branch points for large f and g. This difference is attributed to overestimating the excluded volume at long ranges. As the hybrid method does not show this, localizing just a few segments largely compensates this shortcoming in the SF‐SCF theory.