To measure cuspal deflection and tooth strain, plus marginal leakage and gap formation caused by polymerization shrinkage during direct resin composite restoration of root-filled premolars.Thirty-two first and second maxillary premolars were divided into four groups (n=8). Group 1 had standardised mesio-occlusal-distal (MOD) cavities and served as the control group. Group 2 had endodontic access and root canal treatment through the occlusal floor of the MOD cavity, leaving the axial dentine intact. Group 3 had endodontic access and root canal treatment with the mesial and distal axial dentine removed. Group 4 had endodontic access and root canal treatment with axial dentine removed and a glass ionomer base (GIC). All groups were restored incrementally using a low shrink resin composite. Cuspal deflection was measured using direct current differential transformers (DCDTs), and buccal and palatal strain was measured using strain gauges. Teeth were immersed in 2% methylene blue for 24h, sectioned and scored for leakage and gap formation under light and scanning electron microscopy.Total cuspal deflection was 4.9±1.3μm for the MOD cavity (group 1), 7.8±3.3μm for endodontic access with intact axial dentine (group 2), 12.2±2.6μm for endodontic access without axial dentine (group 3), and 11.1±3.8μm for endodontic access with a GIC base (group 4). Maximum buccal strain was 134±56, 139±61, 251±125, and 183±63μstrain for groups 1–4 respectively, while the maximum palatal strain was 256±215, 184±149, 561±123, 264±87μstrain respectively. All groups showed marginal leakage; however placement of GIC base significantly improved the seal (p=0.007).Cusp deflection and strain increased significantly when axial dentine was removed as part of the endodontic access. Placement of a glass ionomer base significantly reduced tooth strain and marginal leakage. Therefore, a conservative endodontic access and placement of a glass ionomer base are recommended if endodontically treated teeth undergo direct restoration with resin composite.