Context There is concern that deer are shifting forests towards undesirable trajectories, and culling of deer is often advocated to mitigate these impacts. However, culling deer is expensive and sometimes controversial. To reliably ascertain whether such action is beneficial, management-scale experiments are needed. We conducted a management experiment to evaluate the benefits of culling deer in four New Zealand forests. Aims Our experiment tested the predictions that culling deer should increase (1) canopy tree seedling height relative growth rate (SHRGR), and (2) the foliar biomass of understorey species palatable to deer (FBP). Methods Each forest was divided into two 3600-ha areas, with deer culling randomly assigned to one area. Deer abundances were indexed using faecal pellet counts, and forest variables were measured at the start and end of the 8-year experiment. Deer were already at low abundance in one forest and were not culled there. We used structural equation modelling (SEM) with Bayesian variable selection to update our a priori graphical forest-deer model with data from all four forests. Key results Deer abundances were significantly reduced in one forest but increased or did not change in the other two forests in which deer culling occurred. Culling deer did not increase seedling height relative growth rate (SHRGR) or the foliar biomass of understorey species palatable to deer (FBP) in the three areas subject to deer culling compared with the three areas not subject to deer culling. SEM revealed no significant relationships between local-scale deer abundance and either SHRGR or FBP. Rather, tree basal area and the foliar biomass of unpalatable understorey species were important predictors of FBP and SHRGR, respectively, in some forests. Conclusions Our study revealed that culling deer, as currently practiced by New Zealand land managers, did not generate the desired responses in New Zealand forests, possibly due to deer not being culled to sufficiently low densities and/or because forest dynamics and abiotic drivers determined plant growth more than deer. Implications Managers should consider actions other than ineffective deer culling (e.g. creating canopy gaps) to alter the dynamics of New Zealand forests. Alternatively, managers will need to substantially increase culling effort above what is currently practised for this activity to substantially reduce deer populations and thus potentially alter forest dynamics.