Several studies have dealt with the application of the generalized stress intensity factor (GISF) on the failure load prediction of adhesive joints. However, the effect of geometry on the critical value of the GSIF (Hc) is complex and limits its application. Due to the effect of multiple geometrical features and the limited success in the field of adhesive joints, a statistical analysis is a possibility. This paper investigates the impact of different geometrical features on the Hc in single lap adhesive joints. To achieve this, the statistical response surface methodology (RSM) was used to design the experiments and for the statistical analysis. According to the RSM, 31 arrangements of single lap joints were manufactured and tested. In this analysis, the adhesive thickness, adherend thickness, overlap length and also the free length, each in five different levels, were considered. The effect of linear, quadratic and two-way interactions of the geometrical parameters on the Hc and failure load were also studied. It was shown that Hc is most affected by the overlap length. Variation of Hc in term of the free length is by far higher at lower adhesive thicknesses. Also, the effect of substrate thickness on Hc is more considerable for thinner bondlines. The interactions of overlap length/free length and overlap length/adhesive thickness affect the failure load more considerably than the other studied interactions. The effect of free length on the failure load increases with the bondline thickness, while the effect of substrate thickness is stronger for a lower adhesive thickness.