Variable-retention harvesting (VRH) is an approach for sustaining complex structure in managed forests. A criticism of VRH is that ecological benefits may come at a cost of reduced growth of regeneration, due to competition with residual trees. However, the spatial pattern of retention, i.e., dispersed or aggregated, in VRH systems can be manipulated to minimize suppression of regeneration, and resource limitation to regeneration might be mitigated by reduction of woody shrubs. Continued growth of the residual cohort will compensate for growth reduction of regeneration, although this may differ with retention pattern. We examined aboveground whole-stand biomass growth of trees in a VRH experiment in Pinus resinosa forest in Minnesota, USA. Treatments included dispersed retention, aggregated retention, and an uncut control, as well as a shrub treatment (reduced density or ambient). We addressed the following hypotheses: (1) biomass growth of a cohort of planted pine seedlings will be highest with aggregated rather than dispersed retention, (2) biomass growth of the planted seedlings will increase with shrub reduction, and (3) biomass growth of the residual overstory will be higher with dispersed rather than aggregated retention. Aboveground biomass growth of the planted pines ranged from 0.4 kg·ha−1·yr−1 in the overstory-control–ambient-shrub treatment to 23 kg·ha−1·yr−1 in the aggregated-retention–shrub-reduction treatment. The difference between the control and the retention treatments was significant (P < 0.0001), but not between dispersed and aggregated retention (P = 0.97). Thus, our first hypothesis was not supported. In all treatments, biomass growth was significantly higher (>100% increase) with shrub reduction (P = 0.001), supporting our second hypothesis. Biomass growth of residual trees ranged from 2404 kg·ha−1·yr−1 in the uncut-control–ambient-shrub treatment to 1043 kg·ha−1·yr−1 in the aggregated-retention–shrub-reduction treatment. Differences were significant between the control and retention treatments (P = 0.003), and marginally higher with dispersed vs. aggregated retention (P = 0.09), lending support to our third hypothesis. Our results suggest that managers have flexibility in application of VRH and can expect similar stand-level biomass growth of planted regeneration regardless of retention pattern, but somewhat higher stand-level biomass growth of retained trees with dispersed retention.