To assess the quality of dose distributions in real clinical cases for different dimensions of scanned proton pencil beams. The distance between spots (i.e., the grid of delivery) is optimized for each dimension of the pencil beam. The authors vary the σ of the initial Gaussian size of the spot, from σ(x) = σ(y) = 3 mm to σ(x) = σ(y) = 8 mm, to evaluate the impact of the proton beam size on the quality of intensity modulated proton therapy (IMPT) plans. The distance between spots, Δx and Δy, is optimized on the spot plane, ranging from 4 to 12 mm (i.e., each spot size is coupled with the best spot grid resolution). In our Hyperion treatment planning system (TPS), constrained optimization is applied with respect to the organs at risk (OARs), i.e., the optimization tries to satisfy the dose objectives in the planning target volume (PTV) as long as all planning objectives for the OARs are met. Three-field plans for a nasopharynx case, two-field plans for a prostate case, and two-field plans for a malignant pleural mesothelioma case are considered in our analysis. For the head and neck tumor, the best grids (i.e., distance between spots) are 5, 4, 6, 6, and 8 mm for σ = 3, 4, 5, 6, and 8 mm, respectively. σ ≤ 5 mm is required for tumor volumes with low dose and σ ≤ 4 mm for tumor volumes with high dose. For the prostate patient, the best grid is 4, 4, 5, 5, and 5 mm for σ = 3, 4, 5, 6, and 8 mm, respectively. Beams with σ > 3 mm did not satisfy our first clinical requirement that 95% of the prescribed dose is delivered to more than 95% of prostate and proximal seminal vesicles PTV. Our second clinical requirement, to cover the distal seminal vesicles PTV, is satisfied for beams as wide as σ = 6 mm. For the mesothelioma case, the low dose PTV prescription is well respected for all values of σ, while there is loss of high dose PTV coverage for σ > 5 mm. The best grids have a spacing of 6, 7, 8, 9, and 12 mm for σ = 3, 4, 5, 6, and 8 mm, respectively. The maximum acceptable proton pencil beam σ depends on the volume treated, the protocol of delivery, and optimization of the plan. For the clinical cases, protocol and optimization used in this analysis, acceptable σs are ≤ 4 mm for the head and neck tumor, ≤ 3 mm for the prostate tumor and ≤ 6 mm for the malignant pleural mesothelioma. One can apply the same procedure used in this analysis when given a "class" of patients, a σ and a clinical protocol to determine the optimal grid spacing.