The conventional approach to LiDAR-based forest inventory modeling depends on field sample data from fixed-radius plots (FRP). Because FRP sampling is cost intensive, combining variable-radius plot (VRP) sampling and LiDAR data has the potential to improve inventory efficiency. The overarching goal of this study was to evaluate the integration of LiDAR and VRP data. FRP and VRP plots using different basal-area factors (BAF) were colocated in 6 conifer stands near Alberta, Michigan, in the United States. A suite of LiDAR metrics was developed for 24 different resolutions at each plot location, and a number of nonparametric prediction models were evaluated to identify an optimal LiDAR resolution and an optimal scale of VRP to spatially extend the data. An FRP-based model had root mean square error (RMSE) of 31.8 m3 ha−1, whereas the top VRP-based models were somewhat less precise, with RMSE of 38.0 m3 ha−1 and 45.8 m3 ha−1 using BAF 2.06 m2 ha−1 and BAF 2.29 m2 ha−1, respectively. The optimal LiDAR resolution for the VRP data was found to be 18 m for the selected stands, and plot-level estimates based on a model using BAF 2.29 m2 ha−1 were statistically equivalent to the FRP measurements. The use of VRP data shows promise and can substitute for FRP measurements to improve efficiency.